CN113837059B - Patrol vehicle for prompting pedestrians to wear mask in time and control method thereof - Google Patents
Patrol vehicle for prompting pedestrians to wear mask in time and control method thereof Download PDFInfo
- Publication number
- CN113837059B CN113837059B CN202111104530.7A CN202111104530A CN113837059B CN 113837059 B CN113837059 B CN 113837059B CN 202111104530 A CN202111104530 A CN 202111104530A CN 113837059 B CN113837059 B CN 113837059B
- Authority
- CN
- China
- Prior art keywords
- mask
- module
- target
- information
- obstacle
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 50
- 238000007689 inspection Methods 0.000 claims abstract description 74
- 230000033001 locomotion Effects 0.000 claims abstract description 40
- 238000000605 extraction Methods 0.000 claims abstract description 13
- 238000012544 monitoring process Methods 0.000 claims abstract description 4
- 238000001514 detection method Methods 0.000 claims description 72
- 238000010276 construction Methods 0.000 claims description 18
- 230000008569 process Effects 0.000 claims description 18
- 230000007246 mechanism Effects 0.000 claims description 16
- 238000012549 training Methods 0.000 claims description 12
- 238000007781 pre-processing Methods 0.000 claims description 7
- 238000012545 processing Methods 0.000 claims description 7
- 238000013507 mapping Methods 0.000 claims description 6
- 230000000007 visual effect Effects 0.000 claims description 6
- 238000006243 chemical reaction Methods 0.000 claims description 5
- 238000012216 screening Methods 0.000 claims description 5
- 230000009471 action Effects 0.000 claims description 4
- 238000002372 labelling Methods 0.000 claims description 4
- 238000011897 real-time detection Methods 0.000 claims description 4
- 230000001629 suppression Effects 0.000 claims description 4
- 238000005457 optimization Methods 0.000 claims description 3
- 238000012163 sequencing technique Methods 0.000 claims description 3
- 238000013459 approach Methods 0.000 claims description 2
- 230000002265 prevention Effects 0.000 abstract description 17
- 238000013527 convolutional neural network Methods 0.000 abstract description 3
- 230000006399 behavior Effects 0.000 description 5
- 230000006870 function Effects 0.000 description 5
- 238000013528 artificial neural network Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000009223 counseling Methods 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 238000013519 translation Methods 0.000 description 2
- 241000700605 Viruses Species 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000004888 barrier function Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000004806 packaging method and process Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 230000014599 transmission of virus Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/16—Sound input; Sound output
- G06F3/165—Management of the audio stream, e.g. setting of volume, audio stream path
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
- G06Q10/047—Optimisation of routes or paths, e.g. travelling salesman problem
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
- G06T17/05—Geographic models
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- Software Systems (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- General Engineering & Computer Science (AREA)
- Geometry (AREA)
- Economics (AREA)
- Computing Systems (AREA)
- Evolutionary Computation (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Strategic Management (AREA)
- Molecular Biology (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Game Theory and Decision Science (AREA)
- Quality & Reliability (AREA)
- Computer Graphics (AREA)
- Remote Sensing (AREA)
- Entrepreneurship & Innovation (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Development Economics (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Multimedia (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Human Computer Interaction (AREA)
- Traffic Control Systems (AREA)
- Navigation (AREA)
- Image Analysis (AREA)
Abstract
The invention belongs to the technical field of monitoring and early warning security, and particularly relates to a patrol vehicle for prompting pedestrians to wear a mask in time and a control method thereof. According to the indoor public occasion, the indoor grid map is constructed through RGB color images and depth images collected by a depth camera carried by the inspection system; the position of the inspection vehicle relative to the world coordinate system is calculated through feature matching between adjacent pictures, and the self-positioning is completed; by constructing a convolutional neural network model, feature extraction and recognition of whether a pedestrian wears a mask can be completed; the function of detecting autonomous movement of the patrol vehicle to a target is completed through the path planning navigation and obstacle avoidance module; finally, the target following module and the voice module are used for completing the function of the inspection system for prompting pedestrians who do not wear the mask and propaganda work of epidemic prevention knowledge.
Description
Technical Field
The invention belongs to the technical field of monitoring and early warning security, and particularly relates to a patrol vehicle for prompting pedestrians to wear a mask in time and a control method thereof.
Background
Wearing masks in public places becomes an effective means of preventing viral transmission and is also responsible for compliance by each qualified citizen. In places where people are dense, such as railway stations, markets and movie theatres, people are required to actively wear the mask through propaganda means, development of technology at the present stage is combined, people who do not wear the mask are actively discouraged to wear the mask in time through the means of technology, and the possibility of virus transmission is reduced. The intelligent detection inspection vehicle is worn through the indoor mask, real-time detection and persuasion of wearing the mask are carried out on the past personnel in public places, the mask wearing supervision efficiency can be effectively improved, and the intelligent detection inspection vehicle has important significance for epidemic situation prevention and control work.
The design of the intelligent detection inspection vehicle system worn on the indoor mask must have corresponding technical support to complete tasks such as autonomous positioning and map construction, autonomous navigation, target identification and the like. However, through the intelligent detection inspection vehicle, the automatic detection and the active recommendation of the behavior of the unworn mask cannot be effectively put into practical use, and the system for detecting and prompting the behavior of the unworn mask in the public place through the intelligent detection inspection vehicle needs to be developed and put into application as soon as possible to effectively assist the prevention and control work of epidemic situation.
Meanwhile, if the method that relevant staff supervises in public places is adopted, the number of the input staff is relatively large, and unnecessary conflicts with the staff without the mask are easily caused. The intelligent detection inspection vehicle can carry out multi-target detection on personnel based on a deep learning method, obtain the position information of a target without wearing a mask, complete self position estimation through a simultaneous positioning and map building system, and carry out autonomous navigation to the position of the target personnel through an optimal path planning algorithm for prompting. Therefore, labor cost is saved for development of epidemic prevention work, propaganda and education work on epidemic prevention knowledge of people can be completed through the intelligent detection inspection vehicle, awareness of people on safety epidemic prevention is effectively improved, and the intelligent epidemic prevention vehicle has important application prospect.
Disclosure of Invention
The invention aims to provide a patrol vehicle for prompting pedestrians to wear a mask in time.
The aim of the invention is realized by the following technical scheme: the system comprises a map building module, a patrol vehicle positioning module, a mask wearing detection module, a path planning navigation module, an obstacle avoidance module, a target following and voice reminding module;
The map construction module is used for constructing a map according to RGB image information and depth information which are shot and acquired by a depth camera mounted on the inspection vehicle, and is used for establishing a two-dimensional grid map according to the acquired data when cruising for the first time and transmitting the established two-dimensional grid map to the path planning navigation module; the two-dimensional grid map comprises an idle area, an occupied area and an unknown area, wherein the idle area refers to an area through which the patrol vehicle can smoothly pass, the occupied area refers to an area blocked by an obstacle, and the unknown area refers to an area which is not explored by the patrol vehicle;
the inspection vehicle positioning module completes the positioning of the inspection vehicle by the conversion relation among the world coordinate system, the reference coordinate system and the camera coordinate system, and transmits positioning information to the path planning navigation module;
the mask wearing detection module inputs the image information shot and acquired by the depth camera carried on the inspection vehicle into a trained mask detection network, carries out mask detection, acquires the position information of all personnel who do not wear the mask, and transmits the position information of the personnel closest to the mask to the path planning navigation module;
the road strength planning navigation module performs autonomous path planning based on an initial global map established by the map construction module, self-positioning information of the inspection vehicle positioning module and pedestrian position information which is closest to the pedestrian and is transmitted by the mask wearing detection module, and transmits real-time information of surrounding environment of the inspection vehicle and real-time data of self-position to the obstacle avoidance module and the voice reminding module;
The method comprises the steps that surrounding environment information acquired in real time by a depth camera mounted on an obstacle avoidance module inspection vehicle is judged whether a dynamic obstacle is encountered in the overall path planning process, if the dynamic obstacle is detected, position information of the obstacle in a world coordinate system is transmitted to a map building module and a path planning navigation module in real time, planning of a local path is carried out through the path planning navigation module, and an optimal route of the overall path planning is changed to obtain an optimal local path planning after the obstacle is bypassed;
after the inspection vehicle moves to the position of the target, the target following and voice reminding module carries out voice prompt of wearing the mask, and whether the target following mode is started or not is selected through real-time detection of whether the target wears the mask or not; if the target complies with the prompt, actively wearing the mask, and automatically moving the inspection vehicle to the next target; if the target person does not follow the prompt to wear the mask and tries to leave, the target following mode is started, the relative position information of the inspection vehicle and the target is obtained in real time through the depth camera, the target is followed in real time according to the relative position, the voice prompting mode is switched to the voice criticizing mode, and criticizing education is carried out on the target person through the voice module while the target person is followed.
The invention also aims to provide a patrol vehicle control method for prompting pedestrians to wear the mask in time.
The aim of the invention is realized by the following technical scheme: the method comprises the following steps:
step 1: establishing a two-dimensional grid map of the patrol area;
transmitting RGB color picture information and depth information acquired by a depth camera carried by a patrol vehicle to a front-end visual odometer, calculating the movement of cameras between adjacent images and the appearance of a local map according to the estimation of the movement between the adjacent images, and transmitting the camera pose measured by the visual odometer at different moments and the information detected by a loop into a rear-end nonlinear optimization to obtain a globally consistent track and map; judging whether the patrol vehicle reaches the previous position by using loop detection, and if the loop is detected, providing information for back-end processing; the rear end establishes a map model consistent with the task requirements according to the estimated track;
the two-dimensional grid map comprises an idle area, an occupied area and an unknown area, wherein the idle area refers to an area through which the patrol vehicle can smoothly pass, the occupied area refers to an area blocked by an obstacle, and the unknown area refers to an area which is not explored by the patrol vehicle;
Step 2: positioning a patrol vehicle;
based on the established two-dimensional grid map, acquiring the initial position of the patrol vehicle on the two-dimensional grid map; mapping coordinate points in a three-dimensional world to a two-dimensional image plane by a depth camera carried by the inspection vehicle through a pinhole camera model, extracting ORB characteristic points of two acquired adjacent frames of images, extracting a certain threshold characteristic point of each frame of image respectively, measuring the descriptor distance of each characteristic point, sequencing the descriptor distances, and taking the nearest one as a matching point; according to the matched point pairs, the problem of motion estimation between two groups of 3D points is solved by utilizing an ICP method, so that the relative pose of a camera is acquired according to images of two adjacent frames, and the relative pose relationship is mapped to a world coordinate system to obtain real-time position information of the inspection vehicle;
step 3: detecting the wearing of the mask;
transmitting RGB color picture information acquired by a depth camera carried by a patrol vehicle into a trained mask detection network, acquiring position information of pedestrians who do not wear a mask, and selecting the position information of pedestrians with the shortest relative distance as a prompting target;
step 4: path planning navigation and obstacle avoidance;
Establishing a global optimal path and transmitting decision information corresponding to the optimal path to an execution mechanism of the patrol vehicle, driving the patrol vehicle to execute motion operation according to the planned optimal path through linear velocity and angular velocity instructions corresponding to the decision information, and feeding back linear velocity and angular velocity data actually executed by the patrol vehicle by an encoder carried by the execution mechanism in real time, so that the position of the patrol vehicle is continuously optimized until the position approaches to the pose transmitted by the decision information;
in the process that the patrol vehicle moves to a dissuading target through path planning, position information of an obstacle is obtained through a depth camera carried by the outside, and obstacle avoidance is performed according to the position information of the obstacle; firstly, obtaining a distance value of the surrounding environment through the acquired depth information of the picture; comparing the distance value with a pre-stored environment map distance value of the first map construction, judging the distance value as an obstacle if the difference value between the distance value and the pre-stored environment map distance value is smaller than a set threshold value, and continuously detecting the state of the obstacle; judging the movement state of the obstacle according to the distance and the angle of the obstacle detected by the front and the rear times; when the detection system moves to a certain distance from an obstacle, adopting a corresponding obstacle avoidance method to carry out obstacle avoidance movement according to the movement state of the obstacle until the position of a target person is finally reached;
Step 5: target following and voice reminding;
after the inspection vehicle moves to the target position, voice broadcasting reminding is carried out, a target person is reminded of wearing the mask, and real-time monitoring is carried out on the mask; if the target person finishes the action of wearing the mask within a certain time after voice prompt, the target person moves to the position of the next target person; if the target person does not finish wearing the mask after voice prompt, prompting again; if the target personnel try to leave after voice prompt, target following is started, the characteristics of the identity of the followed target are acquired through the depth camera, the following target is tracked in real time after being locked, and meanwhile, the voice prompt is adjusted from a prompting mode to a criticizing mode, and broadcasting is carried out at intervals for a certain time.
The invention may further include:
the method for establishing the two-dimensional grid map in the step 1 specifically comprises the following steps:
step 1.1: setting a map size, an actual distance represented by each grid and a grid threshold;
step 1.2: reading in image data of a depth camera, and preprocessing the data;
step 1.3: initializing to set the whole grid area to black, and representing an unknown area;
step 1.4: the method is based on the principle that the surrounding environment is detected from near to far, the detected area is set to gray, and the detected area represents the range which can be detected by the depth camera; counting the number of times that each grid is projected by the obstacle through the numerical value of the pixel point, and when the number of times is larger than a set threshold value, considering that the corresponding grid map is occupied by the obstacle, and setting the grid map as an occupied area; an area smaller than the threshold value is set as an idle area, which represents that the area can smoothly pass through;
Step 1.5 the step 1.4 is repeatedly executed until the two-dimensional grid map is built.
The training method of the mask detection network in the step 3 is as follows:
firstly, preprocessing image data in the training process, and extracting labeling information from the used image data; sending the image data into a feature extraction network in batches, and obtaining different prediction feature layers generated by each picture through the feature extraction network; generating candidate frames by transmitting different prediction feature layers into a region generation network, generating a series of anchor points by sliding on the prediction feature layers by using a sliding window, performing classification prediction and boundary frame regression prediction on different prediction feature layers by using a convolution layer, applying a prediction result to the generated anchor points to acquire all candidate frame information, and screening the generated candidate frames according to a set threshold value by using non-maximum suppression processing operation; finally, ordering each feature layer of each picture according to the confidence degree from low to high; transmitting candidate frame information generated by the area generating network to the latter half of the network, firstly performing flattening operation, and obtaining final output through two full-connection layers, a classification predictor and a bounding box regression predictor of the network; and reversely transmitting the output, continuously optimizing network parameters through a batch of image data, training to the set iteration times, and obtaining an optimal mask detection network model. .
The invention has the beneficial effects that:
according to the indoor public occasion, the indoor grid map is constructed through RGB color images and depth images collected by a depth camera carried by the inspection system; the position of the inspection vehicle relative to the world coordinate system is calculated through feature matching between adjacent pictures, and the self-positioning is completed; by constructing a convolutional neural network model, feature extraction and recognition of whether a pedestrian wears a mask can be completed; the function of detecting autonomous movement of the patrol vehicle to a target is completed through the path planning navigation and obstacle avoidance module; finally, the inspection system is used for prompting pedestrians who do not wear the mask and propaganda work of epidemic prevention knowledge through the following module and the voice module. The invention can effectively assist relevant staff to complete epidemic prevention work, not only saves labor cost of the epidemic prevention work, but also avoids unnecessary conflict caused by the fact that the staff is dissuaded to wear the mask.
Drawings
Fig. 1 is a system architecture diagram of a patrol vehicle according to the present invention, which encourages pedestrians to wear a mask in time.
Fig. 2 is a flowchart of the implementation of the map construction module in the present invention.
FIG. 3 is a flowchart of the inspection vehicle positioning module according to the present invention.
Fig. 4 is a flowchart of the mask wearing detection module according to the present invention.
Fig. 5 is a flowchart of a path planning navigation module implementation in the present invention.
Fig. 6 is a flowchart of an implementation of the obstacle avoidance module according to the present invention.
FIG. 7 is a flow chart of the implementation of the target following module in the present invention.
FIG. 8 is a flowchart of a voice prompt module according to the present invention.
Detailed Description
The invention is further described below with reference to the accompanying drawings.
The invention provides an automatic inspection vehicle for prompting pedestrians to wear masks in time based on the requirement that the behavior of the masks is monitored in indoor public places in current epidemic situation prevention and control work and a control method thereof. Establishing an indoor grid map through a video stream acquired by a depth camera, and simultaneously completing detection work of personnel who do not wear the mask by utilizing a deep learning network; path planning is carried out through a decision system, and a motion instruction is issued to an executing mechanism to complete autonomous navigation of a target position; meanwhile, the voice broadcasting module is used for prompting pedestrians who do not wear the mask in the indoor public places and conducting epidemic prevention knowledge propaganda, so that propaganda of epidemic prevention knowledge is completed, technical support is provided for prompting pedestrians to wear the mask in the indoor public places, and great significance is provided for epidemic prevention work.
An inspection vehicle for prompting pedestrians to wear a mask in time, comprising: the system comprises a map building module, a patrol vehicle positioning module, a mask wearing detection module, a path planning navigation module, an obstacle avoidance module, a target following module and a voice reminding module;
and a map construction module: and the indoor map is constructed according to the RGB image information and the depth information which are shot and acquired by a depth camera mounted on the inspection vehicle. When the system cruises for the first time, an indoor grid map is built according to the acquired data, wherein the grid map mainly comprises three parts, namely an idle area, an occupied area and an unknown area, the idle area refers to an area through which the system can smoothly pass, the occupied area refers to an area blocked by an obstacle, and the unknown area refers to an area which is not explored by the system yet. And establishing an indoor global map through initial patrol of the system, and transmitting the established two-dimensional grid map to a path planning navigation module.
The inspection vehicle positioning module: and the positioning of the patrol system is completed through the conversion relation among the world coordinate system, the reference coordinate system and the camera coordinate system. The environment picture acquired by the carried depth camera maps three-dimensional space information to a two-dimensional plane through a camera pinhole model principle to acquire a pose transformation relation of a camera coordinate system relative to a world coordinate system, and because the relative position relation between a reference coordinate system and the camera coordinate system is fixed, the position relation of the reference coordinate system relative to the world coordinate system can be calculated through the position relation between the camera coordinate system and the world coordinate system and the position relation between the reference coordinate system relative to the camera coordinate system, so that the real-time positioning function of detecting the inspection vehicle is completed, and the positioning information is transmitted to a path planning navigation module.
Mask wearing detection module: firstly, preprocessing image data in the training process, and extracting labeling information from the used image data; sending the image data into a feature extraction network in batches, and obtaining different prediction feature layers generated by each picture through the feature extraction network; generating candidate frames by transferring different prediction feature layers into a region generation network, mainly using a sliding window of 3*3 to slide on the prediction feature layers to generate a series of anchor points, carrying out classified prediction and bounding box regression prediction on different prediction feature layers through a 1*1 convolution layer, applying a prediction result to the generated anchor points to acquire all candidate frame information, screening the generated candidate frames according to a certain threshold value through non-maximum suppression processing operation, and finally reserving only 2000 candidate frames according to low-to-high confidence degree ordering of each feature layer of each picture; transmitting candidate frame information generated by the area generating network to the latter half of the network, firstly performing flattening operation, and obtaining final output through two full-connection layers, a classification predictor and a bounding box regression predictor of the network; and reversely transmitting the output, continuously optimizing network parameters through a batch of image data, training to the set iteration times, and obtaining the optimal wearing mask detection network model. Image information obtained by shooting by a depth camera mounted on a detection inspection vehicle is detected, pedestrians in an indoor public place are detected through a mask detection network model obtained through training, and whether the pedestrians wear the mask or not is judged through classifying pedestrians who wear the mask. If pedestrians do not wear the mask in the shot pictures, uploading the position information of all people who do not wear the mask to the inspection tour vehicle, judging which person of the people who do not wear the mask is closest to the inspection tour vehicle in relative position through analysis of the position information, and transmitting the position information of the person closest to the inspection tour vehicle to the path planning navigation module, so that a target is provided for the next step of autonomous navigation of the inspection tour vehicle to the designated position for prompting.
Route planning navigation module: and based on the initial global map established by the map construction module, the self-positioning information of the inspection vehicle positioning module and the pedestrian position information closest to the system transmitted by the mask wearing detection module, autonomous path planning is carried out, and moving decision data is issued in real time. Firstly, starting global path planning, namely calculating an optimal path moving from a current position to a target position through a path planning algorithm; then, decision data such as angular speed, linear speed and the like required by the system movement are issued to an executing mechanism through a decision mechanism; the execution structure controls and detects the movement of the inspection vehicle according to the issued instruction, and feeds real movement data detected by the encoder back to the decision-making mechanism in real time to continuously optimize the position, so that the autonomous navigation function of the detection system is realized. And transmitting real-time information of the surrounding environment of the inspection vehicle and real-time data of the position of the inspection vehicle to the obstacle avoidance module and the voice reminding module.
Obstacle avoidance module: based on the surrounding environment information acquired in real time by a depth camera carried by the detection system, judging whether a dynamic obstacle is encountered in the global path planning process of the system. If the existence of the dynamic obstacle is detected, the position information of the obstacle in the world coordinate system is transmitted to a map building module and a path planning navigation module in real time, the path planning navigation module is used for planning a local path, and an optimal route of the global path planning is changed to obtain an optimal local path planning after the obstacle is bypassed.
And a target following module: after the detection system moves to the position of the target, voice prompt of wearing the mask is carried out, and whether the target following mode is started or not is selected through real-time detection of whether the target wears the mask or not. If the target person actively wears the mask according to the prompt, the detection system automatically moves to the next target; if the target person does not follow the prompt to wear the mask and attempts to leave, the target following mode is enabled. And acquiring the relative position information of the inspection vehicle and the target in real time through the depth camera, and following the target in real time according to the relative position. And transmitting the behavior of the target personnel for evading the counseling to the voice prompt module.
The voice reminding module is used for: detecting that the inspection vehicle moves to the position of the target person through the target navigation module and the obstacle avoidance module, and carrying out voice prompt for wearing mask counseling. And carrying out voice reminding on the personnel who do not wear the mask, if the instruction issued by the target following module is received, switching the voice prompting mode to a voice criticizing mode, and carrying out criticizing education on the personnel through the voice module while following the target personnel.
A control method of a patrol vehicle for prompting pedestrians to wear a mask in time comprises the following steps:
(1) And (3) map construction: RGB color picture information and depth information acquired by a depth camera carried by the system are transmitted to a front-end visual odometer, and the movement of the camera between adjacent images and the appearance of a local map are deduced according to the estimation of the movement between the adjacent images; the camera pose measured by the visual odometer at different moments and the information detected by the loop are transmitted to the back-end nonlinear optimization to obtain a globally consistent track and map; using loop detection to judge whether the system reaches the previous position, if the loop is detected, providing information for back-end processing; and the rear end establishes a map model consistent with the task requirements according to the estimated track. And uploading the map model to a patrol vehicle positioning module and a path planning navigation module.
(2) Positioning the inspection vehicle: based on a grid map established by a map construction module, acquiring an initial position of a detection system in the grid map, detecting a depth camera carried by a patrol vehicle, mapping coordinate points (in meters) in a three-dimensional world to a two-dimensional image plane (in pixels) through a pinhole camera model, extracting ORB characteristic points of two acquired adjacent frames of images, extracting certain threshold characteristic points of each frame of images respectively, measuring the descriptor distance of each characteristic point, sequencing the descriptor distance, and taking the nearest one as a matching point. According to the matched point pairs, the problem of motion estimation between two groups of 3D points is solved by utilizing an ICP method, so that the relative pose of a camera is acquired according to images of two adjacent frames, and the relative pose relationship is mapped to a world coordinate system to obtain real-time position information of a detection inspection vehicle. And uploading the position information to a path planning navigation module.
(3) Mask wearing detection step: transmitting RGB color picture information acquired by a depth camera carried by the system into a mask detection network; establishing a network structure of a feature extraction network, and setting weight parameters of an optimal network model trained by the network; carrying out pretreatment operations such as normalization, tensor conversion, size adjustment and the like on each picture in the input port cover detection network; transmitting the preprocessed picture into a prediction model to obtain prediction results such as a target boundary box, a category label, a confidence score and the like; mapping the prediction result to the original image and drawing the category information and the confidence information of the prediction result. The method comprises the steps of detecting video streams through a neural network, acquiring position information of pedestrians without wearing masks, mainly comprising translation information and rotation information of pedestrians under three-dimensional space coordinates, storing the information into a list, selecting pedestrian position information with the shortest relative distance as a prompting target by circularly traversing relative position distance information between all pedestrians without wearing masks in the list and a detection inspection vehicle, and transmitting the pedestrian position information into a path planning navigation module.
(4) A path planning navigation step: packaging and sending initial global map data transmitted by a map construction module, position data of a detected patrol vehicle in a global map transmitted by a patrol vehicle positioning module and target pedestrian position data transmitted by a mask wearing detection module into a path planning navigation initialization function; establishing a global optimal path, transmitting decision information corresponding to the optimal path to an executing mechanism, driving a detection system to execute motion operation according to the planned optimal path through linear velocity and angular velocity instructions corresponding to the decision information, and feeding back linear velocity and angular velocity data actually executed by the system by an encoder carried by the executing mechanism in real time, so that the position of the system is continuously optimized until the pose issued by the decision information is approximated; and finally moving to the target pedestrian position through repeated circulation, and uploading the reaching information to the voice reminding module.
(5) Obstacle avoidance step: and in the process of detecting that the inspection vehicle moves to a target person through path planning, acquiring the position information of the obstacle through a depth camera carried by the outside, and avoiding the obstacle according to the position information of the obstacle. Firstly, obtaining a distance value of the surrounding environment through the acquired depth information of the picture; comparing the distance value with a pre-stored environment map distance value of the first map construction, judging the distance value as an obstacle if the difference value between the distance value and the pre-stored environment map distance value is smaller than a set threshold value, and continuously detecting the state of the obstacle; judging the movement state of the obstacle according to the distance and the angle of the obstacle detected by the front and the rear times; when the detection system moves to the obstacle for a certain distance, a corresponding obstacle avoidance method is adopted to carry out obstacle avoidance movement according to the movement state of the obstacle. Until the position of the target person is finally reached.
(6) The target following step: detecting that the inspection vehicle moves to a target position through the path planning navigation module to carry out voice broadcast reminding, and if the pedestrian still does not wear the mask and tries to leave, starting the target following module. Firstly, acquiring the characteristics of the identity of a followed target through a depth camera; tracking the following target in real time after the following target is locked, and obtaining the relative direction between the following target and the detection patrol vehicle; detecting the locked following target in real time, and calculating to obtain the relative distance between the following target and the detection patrol vehicle; determining a movement route and decision information such as angular speed, linear speed and the like of movement according to the acquired relative direction and relative distance between the inspection vehicle and the pedestrian target; the following of the detection system to the target person is completed by issuing decision information to the execution mechanism.
(7) Voice reminding step: the detection system moves to the position of a target person through the path navigation module, the final position of the target person is determined to be reached through the inspection vehicle positioning module, the voice broadcasting module is executed, the target person is reminded to wear the mask and is monitored in real time, and if the target person finishes the action of wearing the mask within a certain time after voice prompt, the target person moves to the position of the next target person; if the target person does not finish wearing the mask after voice prompt, prompting again; if the target person tries to leave after voice prompt, the information is issued to the target following module, and the voice prompt module is adjusted from a prompting mode to a criticizing mode and broadcasts at intervals of a certain time while the target following module is started.
The invention has the beneficial effects that:
according to the indoor public occasion, the indoor grid map is constructed through RGB color images and depth images collected by a depth camera carried by the inspection system; the position of the inspection vehicle relative to the world coordinate system is calculated through feature matching between adjacent pictures, and the self-positioning is completed; by constructing a convolutional neural network model, feature extraction and recognition of whether a pedestrian wears a mask can be completed; the function of detecting autonomous movement of the patrol vehicle to a target is completed through the path planning navigation and obstacle avoidance module; finally, the inspection system is used for prompting pedestrians who do not wear the mask and propaganda work of epidemic prevention knowledge through the following module and the voice module. The invention can effectively assist relevant staff to complete epidemic prevention work, not only saves labor cost of the epidemic prevention work, but also avoids unnecessary conflict caused by the fact that the staff is dissuaded to wear the mask.
Example 1:
as shown in fig. 1, the invention mainly comprises: the system comprises a map building module, a patrol vehicle positioning module, a mask wearing detection module, a path planning navigation module, an obstacle avoidance module, a target following module and a voice reminding module.
The specific implementation process of each module is as follows:
1 map construction Module
The implementation process of the module is as shown in fig. 2:
(1) Some parameters required to be set for initializing the local map mainly refer to the set map size and the size of grids, namely the actual distance represented by each grid;
(2) Reading in image data of a depth camera carried by the system, and preprocessing the data;
(3) Firstly, setting the whole grid area to be black, wherein the whole grid area represents an unknown area in three modes of the grid map;
(4) The method is based on the principle that the surrounding environment is detected from near to far, and the detected area is set to be gray and represents the range which can be detected by the camera;
(5) Counting the number of times that each grid is projected by the obstacle through the numerical value of the pixel point, when the number of times is larger than a certain threshold value, considering that the corresponding grid map is occupied by the obstacle, setting the grid map as an occupied area, and setting an area smaller than the threshold value as an idle area, wherein the area can smoothly pass through;
(6) And (5) repeating the step (5) until the two-dimensional grid map is built, and uploading the built real-time two-dimensional grid map to the path planning navigation module.
2 inspection vehicle positioning module
The implementation process of the module is as shown in fig. 3:
(1) Initializing a grid map established by a map construction module;
(2) Acquiring an initial position of a detection patrol vehicle on the grid map;
(3) The acquisition system external sensor depth camera maps coordinate points (in meters) in the three-dimensional world to data of a two-dimensional image plane (in pixels) through a pinhole camera model.
(4) Extracting ORB characteristic points of the acquired two adjacent frames of images, and screening the ORB characteristic points according to a certain threshold;
(5) Measuring the descriptor distance of each feature point, sorting, taking the nearest one as a matching point, and matching the feature points between the two images;
(6) According to the matched point pairs, the problem of motion estimation between two groups of 3D points is solved by utilizing an ICP method, and the relative pose of a camera is obtained according to the relative motion of two adjacent frames of images;
(7) Mapping the relative pose relation to a world coordinate system, obtaining real-time position information of the inspection tour bus relative to the world coordinate system, and uploading the position information to a path planning navigation module.
3 mask wearing detection module
The implementation process of the module is as shown in fig. 4:
(1) Initializing parameters required by a mask detection network;
(2) Transmitting RGB color picture information acquired by a depth camera carried by the system into a mask detection network;
(3) Establishing a network structure of a feature extraction network, and setting weight parameters of an optimal network model trained by the network;
(4) Carrying out pretreatment such as normalization, tensor conversion, size adjustment and the like on each picture in the input port cover detection network;
(5) Transmitting the preprocessed picture into a prediction model to obtain prediction results such as a boundary frame, a category label, a confidence score and the like;
(6) Mapping the prediction result to the original image and drawing the category information and the confidence information of the prediction result.
(7) Circularly executing (1) - (6), finishing the detection of the neural network on the input video stream, acquiring the position information of pedestrians who do not wear the mask, mainly comprising translation information and rotation information of pedestrians under three-dimensional space coordinates, and storing the information into a list;
(8) And selecting the pedestrian position information with the shortest relative distance to transmit the pedestrian position information to the path planning navigation module by circularly traversing the relative position distance information of all pedestrians without masks and the detection inspection vehicle in the list.
4-path planning navigation module
The implementation process of the module is as shown in fig. 5:
(1) Initializing parameters required by path planning navigation;
(2) Acquiring initial position information of a system detected by a patrol vehicle positioning module and determining position information of a target pedestrian through a mask detection module;
(3) Planning a global optimal path according to an initial map established by the map construction module;
(4) The detection system moves according to the planned global optimal path, and an encoder corresponding to the execution system uploads the angular speed and linear speed information of the current system in real time;
(5) The information is compared with angular velocity linear velocity information issued by a decision-making mechanism, and negative feedback operation is carried out, so that the pose of the system is continuously optimized, and the expected pose is approximated;
(6) And (4) and (5) circulating until the detection system moves to the target pedestrian position.
5 barrier avoiding module
The implementation process of the module is shown in fig. 6:
(1) Initializing parameters required by obstacle avoidance;
(2) Obtaining a distance value of the surrounding environment through the obtained depth information of the picture;
(3) Comparing the distance value with an environment map distance value pre-stored in the first map construction;
(4) If the difference value between the distance value and the pre-stored environmental map distance value is larger than the set threshold value, judging the obstacle as the obstacle, and continuously detecting and refreshing the state of the obstacle;
(5) Judging the movement state of the obstacle according to the distance and angle of the obstacle detected by the front and rear times;
(6) When the detection system moves to an obstacle for a certain distance, adopting a corresponding obstacle avoidance method to carry out obstacle avoidance movement according to the movement state of the obstacle;
(7) And (3) circulating the steps (2) - (6) until the target personnel are finally and autonomously moved to the position.
6 target following module
The implementation process of the module is shown in fig. 7:
(1) Initializing parameters required by target following;
(2) After reminding a certain time, the voice broadcast judges whether the pedestrian wears the mask, and if the pedestrian still does not wear the mask and tries to leave, the target following module is started;
(3) Acquiring characteristics of a following target person through a depth camera;
(4) Tracking the following target in real time after the following target is locked, and obtaining the relative direction between the following target and the detection patrol vehicle;
(5) Detecting the locked following target in real time, and calculating to obtain the relative distance between the following target and the detection patrol vehicle;
(6) Determining a movement route, angular speed, linear speed and other decision information of movement according to the acquired relative direction and relative distance between the inspection vehicle and the pedestrian target;
(7) And finishing detecting the following behavior of the patrol vehicle to the target person by transmitting the decision information to an executing mechanism.
7 voice reminding module
The implementation process of the module is as shown in fig. 8:
(1) Initializing parameters required by voice reminding;
(2) Judging whether the specified position is reached or not through the inspection vehicle positioning module, triggering voice broadcasting if the specified position is reached, and not triggering if the specified region is not reached;
(3) The target is monitored in real time, and if the target personnel finish the action of wearing the mask within a certain time after voice prompt, the target personnel move to the position of the next target personnel;
(4) If the target person does not finish wearing the mask after voice prompt, prompting again;
(5) If the target person tries to leave after voice prompt, the information is issued to the target following module;
the voice prompt module is adjusted from the prompting mode to the criticizing mode and broadcasts once at intervals.
The above description is only of the preferred embodiments of the present invention and is not intended to limit the present invention, but various modifications and variations can be made to the present invention by those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (3)
1. An inspection vehicle for prompting pedestrians to wear a mask in time, which is characterized in that: the system comprises a map building module, a patrol vehicle positioning module, a mask wearing detection module, a path planning navigation module, an obstacle avoidance module, a target following and voice reminding module;
The map construction module is used for constructing a map according to RGB image information and depth information which are shot and acquired by a depth camera mounted on the inspection vehicle, and is used for establishing a two-dimensional grid map according to the acquired data when cruising for the first time and transmitting the established two-dimensional grid map to the path planning navigation module; the two-dimensional grid map comprises an idle area, an occupied area and an unknown area, wherein the idle area refers to an area through which the patrol vehicle can smoothly pass, the occupied area refers to an area blocked by an obstacle, and the unknown area refers to an area which is not explored by the patrol vehicle;
the inspection vehicle positioning module completes the positioning of the inspection vehicle by the conversion relation among the world coordinate system, the reference coordinate system and the camera coordinate system, and transmits positioning information to the path planning navigation module;
the mask wearing detection module inputs the image information shot and acquired by the depth camera carried on the inspection vehicle into a trained mask detection network, carries out mask detection, acquires the position information of all personnel who do not wear the mask, and transmits the position information of the personnel closest to the mask to the path planning navigation module;
the training method of the mask detection network comprises the following steps: firstly, preprocessing image data in the training process, and extracting labeling information from the used image data; sending the image data into a feature extraction network in batches, and obtaining different prediction feature layers generated by each picture through the feature extraction network; generating candidate frames by transmitting different prediction feature layers into a region generation network, generating a series of anchor points by sliding on the prediction feature layers by using a sliding window, performing classification prediction and boundary frame regression prediction on different prediction feature layers by using a convolution layer, applying a prediction result to the generated anchor points to acquire all candidate frame information, and screening the generated candidate frames according to a set threshold value by using non-maximum suppression processing operation; finally, ordering each feature layer of each picture according to the confidence degree from low to high; transmitting candidate frame information generated by the area generating network to the latter half of the network, firstly performing flattening operation, and obtaining final output through two full-connection layers, a classification predictor and a bounding box regression predictor of the network; reversely transmitting the output, continuously optimizing network parameters through a batch of image data, training to the set iteration times, and obtaining an optimal mask detection network model;
The path planning navigation module performs autonomous path planning based on an initial global map established by the map construction module, self-positioning information of the inspection vehicle positioning module and pedestrian position information closest to the pedestrian position information transmitted by the mask wearing detection module, and transmits real-time information of surrounding environment of the inspection vehicle and real-time data of self-position to the obstacle avoidance module and the voice reminding module;
the path planning navigation module firstly performs global path planning, calculates an optimal path from a current position to a target position through a path planning algorithm, and then issues angular speed and linear speed decision data required by system movement to the execution mechanism through the decision mechanism; the execution structure controls and detects the movement of the inspection vehicle according to the issued instruction, and feeds real movement data detected by the encoder back to the decision-making mechanism in real time to continuously optimize the position, so that the autonomous navigation function of the detection system is realized;
the method comprises the steps that surrounding environment information acquired in real time by a depth camera mounted on an obstacle avoidance module inspection vehicle is judged whether a dynamic obstacle is encountered in the overall path planning process, if the dynamic obstacle is detected, position information of the obstacle in a world coordinate system is transmitted to a map building module and a path planning navigation module in real time, planning of a local path is carried out through the path planning navigation module, and an optimal route of the overall path planning is changed to obtain an optimal local path planning after the obstacle is bypassed;
After the inspection vehicle moves to the position of the target, the target following and voice reminding module carries out voice prompt of wearing the mask, and whether the target following mode is started or not is selected through real-time detection of whether the target wears the mask or not; if the target complies with the prompt, actively wearing the mask, and automatically moving the inspection vehicle to the next target; if the target person does not follow the prompt to wear the mask and tries to leave, the target following mode is started, the relative position information of the inspection vehicle and the target is obtained in real time through the depth camera, the target is followed in real time according to the relative position, the voice prompting mode is switched to the voice criticizing mode, and criticizing education is carried out on the target person through the voice module while the target person is followed.
2. A patrol vehicle control method for prompting pedestrians to wear a mask in time is characterized by comprising the following steps:
step 1: establishing a two-dimensional grid map of the patrol area;
transmitting RGB color picture information and depth information acquired by a depth camera carried by a patrol vehicle to a front-end visual odometer, calculating the movement of cameras between adjacent images and the appearance of a local map according to the estimation of the movement between the adjacent images, and transmitting the camera pose measured by the visual odometer at different moments and the information detected by a loop into a rear-end nonlinear optimization to obtain a globally consistent track and map; judging whether the patrol vehicle reaches the previous position by using loop detection, and if the loop is detected, providing information for back-end processing; the rear end establishes a map model consistent with the task requirements according to the estimated track;
The two-dimensional grid map comprises an idle area, an occupied area and an unknown area, wherein the idle area refers to an area through which the patrol vehicle can smoothly pass, the occupied area refers to an area blocked by an obstacle, and the unknown area refers to an area which is not explored by the patrol vehicle;
step 2: positioning a patrol vehicle;
based on the established two-dimensional grid map, acquiring the initial position of the patrol vehicle on the two-dimensional grid map; mapping coordinate points in a three-dimensional world to a two-dimensional image plane by a depth camera carried by the inspection vehicle through a pinhole camera model, extracting ORB characteristic points of two acquired adjacent frames of images, extracting a certain threshold characteristic point of each frame of image respectively, measuring the descriptor distance of each characteristic point, sequencing the descriptor distances, and taking the nearest one as a matching point; according to the matched point pairs, the problem of motion estimation between two groups of 3D points is solved by utilizing an ICP method, so that the relative pose of a camera is acquired according to images of two adjacent frames, and the relative pose relationship is mapped to a world coordinate system to obtain real-time position information of the inspection vehicle;
step 3: detecting the wearing of the mask;
transmitting RGB color picture information acquired by a depth camera carried by a patrol vehicle into a trained mask detection network, acquiring position information of pedestrians who do not wear a mask, and selecting the position information of pedestrians with the shortest relative distance as a prompting target;
The training method of the mask detection network comprises the following steps:
firstly, preprocessing image data in the training process, and extracting labeling information from the used image data; sending the image data into a feature extraction network in batches, and obtaining different prediction feature layers generated by each picture through the feature extraction network; generating candidate frames by transmitting different prediction feature layers into a region generation network, generating a series of anchor points by sliding on the prediction feature layers by using a sliding window, performing classification prediction and boundary frame regression prediction on different prediction feature layers by using a convolution layer, applying a prediction result to the generated anchor points to acquire all candidate frame information, and screening the generated candidate frames according to a set threshold value by using non-maximum suppression processing operation; finally, ordering each feature layer of each picture according to the confidence degree from low to high; transmitting candidate frame information generated by the area generating network to the latter half of the network, firstly performing flattening operation, and obtaining final output through two full-connection layers, a classification predictor and a bounding box regression predictor of the network; reversely transmitting the output, continuously optimizing network parameters through a batch of image data, training to the set iteration times, and obtaining an optimal mask detection network model;
Step 4: path planning navigation and obstacle avoidance;
establishing a global optimal path and transmitting decision information corresponding to the optimal path to an execution mechanism of the patrol vehicle, driving the patrol vehicle to execute motion operation according to the planned optimal path through linear velocity and angular velocity instructions corresponding to the decision information, and feeding back linear velocity and angular velocity data actually executed by the patrol vehicle by an encoder carried by the execution mechanism in real time, so that the position of the patrol vehicle is continuously optimized until the position approaches to the pose transmitted by the decision information;
in the process that the patrol vehicle moves to a dissuading target through path planning, position information of an obstacle is obtained through a depth camera carried by the outside, and obstacle avoidance is performed according to the position information of the obstacle; firstly, obtaining a distance value of the surrounding environment through the acquired depth information of the picture; comparing the distance value with a pre-stored environment map distance value of the first map construction, judging the distance value as an obstacle if the difference value between the distance value and the pre-stored environment map distance value is smaller than a set threshold value, and continuously detecting the state of the obstacle; judging the movement state of the obstacle according to the distance and the angle of the obstacle detected by the front and the rear times; when the detection system moves to a certain distance from an obstacle, adopting a corresponding obstacle avoidance method to carry out obstacle avoidance movement according to the movement state of the obstacle until the position of a target person is finally reached;
Step 5: target following and voice reminding;
after the inspection vehicle moves to the target position, voice broadcasting reminding is carried out, a target person is reminded of wearing the mask, and real-time monitoring is carried out on the mask; if the target person finishes the action of wearing the mask within a certain time after voice prompt, the target person moves to the position of the next target person; if the target person does not finish wearing the mask after voice prompt, prompting again; if the target personnel try to leave after voice prompt, target following is started, the characteristics of the identity of the followed target are acquired through the depth camera, the following target is tracked in real time after being locked, and meanwhile, the voice prompt is adjusted from a prompting mode to a criticizing mode, and broadcasting is carried out at intervals for a certain time.
3. The inspection vehicle control method for prompting pedestrians to wear the mask in time according to claim 2, wherein the method comprises the following steps: the method for establishing the two-dimensional grid map in the step 1 specifically comprises the following steps:
step 1.1: setting a map size, an actual distance represented by each grid and a grid threshold;
step 1.2: reading in image data of a depth camera, and preprocessing the data;
step 1.3: initializing to set the whole grid area to black, and representing an unknown area;
Step 1.4: the method is based on the principle that the surrounding environment is detected from near to far, the detected area is set to gray, and the detected area represents the range which can be detected by the depth camera; counting the number of times that each grid is projected by the obstacle through the numerical value of the pixel point, and when the number of times is larger than a set threshold value, considering that the corresponding grid map is occupied by the obstacle, and setting the grid map as an occupied area; an area smaller than the threshold value is set as an idle area, which represents that the area can smoothly pass through;
step 1.5 the step 1.4 is repeatedly executed until the two-dimensional grid map is built.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111104530.7A CN113837059B (en) | 2021-09-22 | 2021-09-22 | Patrol vehicle for prompting pedestrians to wear mask in time and control method thereof |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111104530.7A CN113837059B (en) | 2021-09-22 | 2021-09-22 | Patrol vehicle for prompting pedestrians to wear mask in time and control method thereof |
Publications (2)
Publication Number | Publication Date |
---|---|
CN113837059A CN113837059A (en) | 2021-12-24 |
CN113837059B true CN113837059B (en) | 2023-12-19 |
Family
ID=78960152
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202111104530.7A Active CN113837059B (en) | 2021-09-22 | 2021-09-22 | Patrol vehicle for prompting pedestrians to wear mask in time and control method thereof |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113837059B (en) |
Families Citing this family (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114821388A (en) * | 2022-03-10 | 2022-07-29 | 慧之安信息技术股份有限公司 | Indoor mask detection method and device based on edge calculation |
CN116164770B (en) * | 2023-04-23 | 2023-07-25 | 禾多科技(北京)有限公司 | Path planning method, path planning device, electronic equipment and computer readable medium |
CN117011231B (en) * | 2023-06-27 | 2024-04-09 | 盐城工学院 | Strip steel surface defect detection method and system based on improved YOLOv5 |
CN117475115B (en) * | 2023-11-11 | 2024-06-21 | 华中师范大学 | Working method of path guiding system in virtual-real fusion environment |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107167139A (en) * | 2017-05-24 | 2017-09-15 | 广东工业大学 | A kind of Intelligent Mobile Robot vision positioning air navigation aid and system |
CN111860230A (en) * | 2020-07-03 | 2020-10-30 | 北京航空航天大学 | Automatic detection system and method based on behavior of video monitoring personnel not wearing mask |
CN113110457A (en) * | 2021-04-19 | 2021-07-13 | 杭州视熵科技有限公司 | Autonomous coverage inspection method for intelligent robot in indoor complex dynamic environment |
CN113175996A (en) * | 2021-05-19 | 2021-07-27 | 云从科技集团股份有限公司 | Temperature monitoring method and device, computer readable storage medium and computer equipment |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110703747B (en) * | 2019-10-09 | 2021-08-03 | 武汉大学 | Robot autonomous exploration method based on simplified generalized Voronoi diagram |
-
2021
- 2021-09-22 CN CN202111104530.7A patent/CN113837059B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107167139A (en) * | 2017-05-24 | 2017-09-15 | 广东工业大学 | A kind of Intelligent Mobile Robot vision positioning air navigation aid and system |
CN111860230A (en) * | 2020-07-03 | 2020-10-30 | 北京航空航天大学 | Automatic detection system and method based on behavior of video monitoring personnel not wearing mask |
CN113110457A (en) * | 2021-04-19 | 2021-07-13 | 杭州视熵科技有限公司 | Autonomous coverage inspection method for intelligent robot in indoor complex dynamic environment |
CN113175996A (en) * | 2021-05-19 | 2021-07-27 | 云从科技集团股份有限公司 | Temperature monitoring method and device, computer readable storage medium and computer equipment |
Non-Patent Citations (1)
Title |
---|
基于机器视觉的室内无轨巡检机器人导航***;汤义勤;高彦波;邹宏亮;叶建军;曾林;;自动化与仪表(第08期);全文 * |
Also Published As
Publication number | Publication date |
---|---|
CN113837059A (en) | 2021-12-24 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN113837059B (en) | Patrol vehicle for prompting pedestrians to wear mask in time and control method thereof | |
Zhang et al. | Vehicle tracking and speed estimation from roadside lidar | |
Bascetta et al. | Towards safe human-robot interaction in robotic cells: an approach based on visual tracking and intention estimation | |
US11475671B2 (en) | Multiple robots assisted surveillance system | |
US20230222671A1 (en) | System for predicting near future location of object | |
US20150262018A1 (en) | Detecting moving vehicles | |
CN112541449A (en) | Pedestrian trajectory prediction method based on unmanned aerial vehicle aerial photography view angle | |
Suleymanov et al. | Online inference and detection of curbs in partially occluded scenes with sparse lidar | |
CN105014675A (en) | Intelligent mobile robot visual navigation system and method in narrow space | |
Nguyen et al. | Confidence-aware pedestrian tracking using a stereo camera | |
CN114905512A (en) | Panoramic tracking and obstacle avoidance method and system for intelligent inspection robot | |
CN110824495B (en) | Laser radar-based drosophila visual inspired three-dimensional moving target detection method | |
Aryal | Object detection, classification, and tracking for autonomous vehicle | |
Mannion | Vulnerable road user detection: state-of-the-art and open challenges | |
Xanthidis et al. | Towards mapping of underwater structures by a team of autonomous underwater vehicles | |
Adachi et al. | Turning at intersections using virtual lidar signals obtained from a segmentation result | |
Kim et al. | Traffic Accident Detection Based on Ego Motion and Object Tracking | |
Fries et al. | Real-time unsupervised feature model generation for a vehicle following system | |
Zhao et al. | A laser-scanner-based approach toward driving safety and traffic data collection | |
EP4089649A1 (en) | Neuromorphic cameras for aircraft | |
Llorca et al. | Assistive pedestrian crossings by means of stereo localization and rfid anonymous disability identification | |
Budzan | Fusion of visual and range images for object extraction | |
CN114800615A (en) | Robot real-time scheduling system and method based on multi-source perception | |
Liu et al. | A Drone Patrol System for Target Object Counting and Geolocalization | |
Illmann et al. | Statistical recognition of motion patterns |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |