CN212445254U

CN212445254U - Autonomous picking robot based on visual algorithm

Info

Publication number: CN212445254U
Application number: CN202021077138.9U
Authority: CN
Inventors: 陈南款; 曾小娅; 陈巧兰
Original assignee: Shenzhen Fund Technology Co ltd
Current assignee: Shenzhen Fund Technology Co ltd
Priority date: 2020-06-11
Filing date: 2020-06-11
Publication date: 2021-02-02
Anticipated expiration: 2030-06-11

Abstract

The utility model relates to an autonomous picking robot based on visual algorithm, which comprises a control mechanism, a supporting mechanism, a driving mechanism, a picking mechanism and an anti-collision detection mechanism; the control mechanism comprises an MCU; the supporting mechanism comprises a chassis and a framework thereof; the driving mechanism comprises a driving wheel and a tail universal wheel; the picking mechanism comprises a picking driving motor, a picking brush roller, a ball conveying groove and a tail box; anticollision detection mechanism is including setting up in crashproof strip sensor, the ultrasonic sensor of locomotive. The utility model discloses have very strong practicality and security, pick up efficiency and the degree of accuracy than higher. The utility model uses the camera support which can be lifted and the wide-angle local camera, which is compatible with the portability and the larger visual range; the ball picking brush has the advantages that the ball picking brush is high in practicability and safe, the brush can protect a ball body, the brush cannot be damaged even if obstacles appear, and a single ball or multiple balls can be quickly picked into the tail box.

Description

Autonomous picking robot based on visual algorithm

Technical Field

The utility model belongs to the robot field, concretely relates to independently pick up robot based on vision algorithm.

Background

With the improvement of living standard, more and more ball enthusiasts are provided at present, and the ball sports have good body-building effect from table tennis, badminton, tennis and golf. In these ball games, especially tennis, we find that a large part of the time is consumed in the ball picking process, and the mechanical action of ball picking is very tedious work, which greatly reduces the efficiency and experience effect of the ball games, so that the player can be released from the tedious ball picking process, and the ball picking machine has very high practical value. Similar problems exist with other object pick-ups, such as refuse collection, loose object pick-ups, etc.

With the advent of the artificial intelligence era, people are constantly trying to replace manual work with robots to accomplish repeated work with machines. Patent application No. 201821674958.9, which proposes an autonomous picking robot for automatically picking table tennis balls; patent application No. 201720658716.X, which proposes a binocular vision automatic ball picking robot; patent application No. 201710345365.1, proposes a tennis ball picking robot; patent application No. 201710131530.3, a ball picking robot based on opencv image recognition is proposed.

At present, ball picking equipment in a ball hall mainly needs manual operation, but more and more intelligent modification schemes are being introduced. However, the solutions generally have the defects of poor practicability, complex structural design or scene constraints, and do not consider the factors such as safety. If patent CN200810188136.4 mentions a scheme for global positioning of a robot using a top view of a camera, but the scheme is not applicable to indoor scenes and outdoor scenes, and patent CN201621301712.8 mentions that global positioning is performed using an unmanned airborne camera, on one hand, the scheme has high operation difficulty and on the other hand, there is a safety hazard. For example, in patent 201811351346.0, patent 201720882172.5, patent 201710345365.1 and the like, the baffle brushes for picking up the balls leak outwards, so that a relatively large risk potential exists.

In the aspect of avoiding the barrier, the current ball picking-up equipment mainly uses ultrasonic waves or infrared barriers, but in practice, the net fence in an outdoor court is not an entity, and the ultrasonic waves possibly lose effectiveness, so that more sensors are needed to comprehensively improve the anti-collision capacity.

SUMMERY OF THE UTILITY MODEL

To the above problem, the utility model provides an adopt multisensor to synthesize keep away the barrier, can navigate fast to the target of longer distance and pick up, the autonomous picking robot based on vision algorithm that the security performance is high.

In order to solve the above problems, the technical scheme adopted by the application is as follows: an autonomous picking robot based on a visual algorithm comprises a control mechanism, a supporting mechanism, a driving mechanism, a picking mechanism and an anti-collision detection mechanism; the control mechanism comprises an MCU; the supporting mechanism comprises a chassis and a framework thereof; the driving mechanism comprises two parallel driving wheels arranged on the front side of the chassis and a tail universal wheel arranged on the rear side of the chassis, and the tail universal wheel is arranged in the middle of the tail of the chassis and serves as a bearing wheel of the tail. The two driving wheels are connected with two speed reduction driving motors with the encoders, the speed reduction driving motors with the encoders are connected with the MCU through the motor drivers and the arduino controller, and the speed reduction driving motors with the encoders provide rotating power for the driving wheels and provide coded data for the MCU; the picking mechanism comprises a picking driving motor, a picking brush roller, a ball transfer groove and a tail box, wherein the picking driving motor is connected with the MCU through a motor driver and an arduino controller, the picking driving motor provides rotation power for the picking brush roller, when a target object enters the rotation range of the brush, the picking brush roller brings the target object into the ball transfer groove, and the target object is conveyed into the tail box through the ball transfer groove, so that the target object is conveniently conveyed, and the ball transfer groove is arranged at a certain angle with the chassis, which can be 45 degrees; anticollision detection mechanism sets up in the local wide angle camera that can tensile camera support and crashproof strip sensor, ultrasonic sensor and local wide angle camera all are connected with MCU including setting up in crashproof strip sensor, the ultrasonic sensor of locomotive.

Preferably, the camera support is provided with a multi-section stretching structure, and a button for controlling the height of the stretching structure is arranged at the top of the camera support. The method aims at the problems that the visual range is small and the relative position and angle between a picked target and a robot cannot be calculated due to the fact that the ball picking camera is low in height. The utility model adopts the camera support which can be pulled up, the camera support can be retracted into the shell when the robot is carried, the camera can be pulled out when picking up the ball, and the camera can reach the height of the ball child; meanwhile, the wide-angle local camera is used for improving the visual field range of the robot, so that the picking robot can acquire accurate navigation coordinates, a target at a longer distance can be quickly navigated and picked, and the picking speed can be improved.

Preferably, the chassis is a concave PVC bottom plate, and the concave chassis can enable balls to enter the robot tail box along the groove of the chassis.

Preferably, a supporting frame is arranged on the upper side face of the chassis, and a picking driving motor and a picking brush roller are hung on the supporting frame; tensile camera support set up in the support frame top.

Preferably, the MCU is connected with a nine-axis degree of freedom IMU sensor module.

Preferably, four groups of brushes are arranged on the picking brush roller, the brushes have larger radius, two rows of brush wires are arranged in a staggered mode in each group of brushes, and the brush wires are made of Ninong materials.

Preferably, a global wide-angle camera connected with the MCU through WIFI is arranged in the court, and positioning markers are arranged at the head and the tail of the robot.

Preferably, a rechargeable battery can be selected to provide a power source for the whole robot.

Preferably, a voltage display is arranged at the position of the vehicle head, and the current voltage and the current electric quantity are displayed in real time.

The utility model provides an automatic scheme of picking up with very strong practicality contains the mechanical structure device of picking up, picks up target detection and navigation and picks up, multi-sensor synthesizes and keeps away functions such as the robot global positioning of barrier, global camera, and human-computer interaction is convenient simple simultaneously, compares current scheme of picking up, the utility model discloses have very strong practicality and security, pick up efficiency and the degree of accuracy than higher.

The utility model uses the camera support which can be lifted and the wide-angle local camera, which is compatible with the portability and the larger visual range; the ball picking brush has the advantages that the ball picking brush is high in practicability and safe, the brush can protect a ball body, the brush cannot be damaged even if obstacles appear, and a single ball or multiple balls can be quickly picked into the tail box. Through the configuration optimization above, the utility model discloses a pick up the target of picking up that the robot can be quick and navigate and pick up, pick up efficiency than higher.

Drawings

FIG. 1 is a schematic view of a pick-up robot structure;

fig. 2 is a perspective view of the pick-up robot;

fig. 3 is a schematic view of the internal structure of the picking-up robot;

fig. 4 is a schematic view of a pick-up brush roller.

1. Chassis, 2, drive wheel, 3, afterbody universal wheel, 4, pick up the brush roller, 5, camera support, 6, local wide angle camera, 7, locomotive, 8, boot.

Detailed Description

Example one

As shown in fig. 1, an autonomous picking robot based on a vision algorithm includes a control mechanism, a supporting mechanism, a driving mechanism, a picking mechanism and an anti-collision detection mechanism; the control mechanism comprises an MCU; the supporting mechanism comprises a chassis 1 and a framework thereof; the driving mechanism comprises two parallel driving wheels 2 arranged on the front side of the chassis 1 and a tail universal wheel 3 arranged on the rear side of the chassis, and the tail universal wheel 3 is arranged in the middle of the tail of the chassis 1 and serves as a bearing wheel of the tail. The two driving wheels 2 are connected with two speed reduction driving motors with encoders, the speed reduction driving motors with the encoders are connected with the MCU through a motor driver and an arduino controller, and the speed reduction driving motors with the encoders provide rotating power for the driving wheels 2 and provide encoded data for the MCU; the picking mechanism comprises a picking driving motor, a picking hairbrush roller 4, a ball transfer groove and a tail box 8, the picking driving motor is connected with the MCU through a motor driver and an arduino controller, the picking driving motor provides rotation power for the picking hairbrush roller 4, when a target object enters the rotation range of the hairbrush, the picking hairbrush roller 4 brings the target object into the ball transfer groove, and the target object is conveyed into the tail box 8 through the ball transfer groove, so that the target object is conveniently conveyed, the ball transfer groove is arranged at a certain angle with the chassis 1, and the angle can be 45 degrees; anticollision detection mechanism is including setting up in crashproof strip sensor, the ultrasonic sensor of locomotive 7. Set up in local wide angle camera 6, crashproof strip sensor, the ultrasonic sensor that can tensile camera support 5 all to be connected with MCU.

The method aims at the problems that the visual range is small and the relative position and angle between a picked target and a robot cannot be calculated due to the fact that the ball picking camera is low in height. The utility model adopts the camera support 5 which can be pulled up, the camera support 5 can be pulled out when the robot is carried and the robot is retracted into the shell, so as to ensure that the camera can reach the height of the ball child; meanwhile, the wide-angle local camera is used for improving the visual field range of the robot, so that the picking robot can acquire accurate navigation coordinates, a target at a longer distance can be quickly navigated and picked, and the picking speed can be improved.

Preferably, the chassis 1 is a concave PVC base plate, and the concave chassis 1 can allow balls to enter the robot tail box 8 along the grooves of the chassis. The front end of the chassis is provided with the U-shaped notch, so that a target object can enter a brush picking range along the U-shaped edge, and the tail part of the chassis is semicircular, so that the user can conveniently fall off when encountering obstacles; the framework is made of alloy materials, so that the supporting mechanism has strong bearing capacity, and excessive weight is not added to the supporting mechanism.

Preferably, a supporting frame is arranged on the upper side face of the chassis, and a picking driving motor and a picking brush roller 4 are hung on the supporting frame; tensile camera support 5 set up in the support frame top.

Preferably, the MCU is connected with a nine-axis degree of freedom IMU sensor module. The nine-axis degree of freedom IMU sensor module comprises a three-axis gyroscope, three-axis acceleration and a three-axis magnetic field, and is used together with an encoder of a motor, so that the positioning precision is improved.

As shown in fig. 4, four groups of brushes are arranged on the picking brush roller 4, the brushes have a larger radius, two rows of brush wires are arranged in a staggered manner in each group of brushes, and the brush wires are made of ninon material.

Preferably, a global wide-angle camera connected with the MCU through WIFI is arranged in the court, and positioning markers are arranged at the head and the tail of the robot. The utility model provides a set up the scheme of two fisheye cameras at court border, single fisheye camera is responsible for the robot location that corresponds half a court at place, marks the camera through the discernment in court marking line, sets up the marker at robot head and afterbody, position that can the real-time supervision robot to guide the robot to carry out quick global search.

Claims

1. An autonomous picking robot based on a vision algorithm is characterized in that: the anti-collision device comprises a control mechanism, a supporting mechanism, a driving mechanism, a picking mechanism and an anti-collision detection mechanism; the control mechanism comprises an MCU; the supporting mechanism comprises a chassis and a framework thereof; the driving mechanism comprises a driving wheel arranged on the front side of the chassis and a tail universal wheel arranged on the rear side of the chassis, the driving wheel is connected with a speed reduction driving motor with a coder, the speed reduction driving motor with the coder is connected with the MCU through a motor driver and an arduino controller, and the speed reduction driving motor with the coder provides rotating power for the driving wheel and provides coded data for the MCU; the picking mechanism comprises a picking driving motor, a picking brush roller, a ball transfer groove and a tail box, the picking driving motor is connected with a motor driver and is connected with the MCU through an arduino controller, the picking driving motor provides rotating power for the picking brush roller, the picking brush roller brings a target object into the ball transfer groove, and the target object is conveyed into the tail box through the ball transfer groove; the anti-collision detection mechanism comprises an anti-collision strip sensor and an ultrasonic sensor which are arranged on the vehicle head; set up in the local wide angle camera that can tensile camera support with crashproof strip sensor, ultrasonic sensor all be connected with MCU.

2. The vision algorithm-based autonomous pick-up robot of claim 1, wherein: the camera support is of a multi-section stretching structure, and a button for controlling the height of the stretching structure is arranged at the top of the camera support.

3. The vision algorithm-based autonomous pick-up robot of claim 1, wherein: the chassis is a concave PVC bottom plate, the front end of the chassis is a U-shaped notch, and the tail of the chassis is semicircular; the framework is made of alloy materials.

4. The vision algorithm-based autonomous pick-up robot of claim 1, wherein: a supporting frame is arranged on the upper side face of the chassis, and a picking driving motor and a picking hairbrush roller are hung on the supporting frame; tensile camera support set up in the support frame top.

5. The vision algorithm-based autonomous pick-up robot of claim 1, wherein: the picking brush roller is provided with four groups of brushes, each group of brushes is provided with two rows of brush wires in a staggered arrangement mode, and the brush wires are made of Ninong materials.

6. The vision algorithm-based autonomous pick-up robot of claim 1, wherein: the MCU is connected with a nine-axis degree of freedom IMU sensor module.

7. The vision algorithm-based autonomous pick-up robot of claim 1, wherein: the global wide-angle camera connected with the MCU through the WIFI is arranged in the court, positioning markers are arranged at the head and the tail of the robot, and the position of the robot in the court can be positioned in real time through calibration based on court marker lines.