Disclosure of Invention
The invention aims to provide a coal mine fully-mechanized mining face inspection unmanned vehicle based on SLAM technology and an autonomous inspection method, which are high in positioning accuracy and small in error.
The technical scheme of the invention is that the coal mine fully-mechanized mining face inspection unmanned vehicle based on the SLAM technology comprises a binocular camera, a protective frame, a chassis and an explosion-proof box, and is characterized in that: the unmanned vehicle is also provided with an intelligent module and a sensor module; the intelligent module comprises a coal mine environment image processing module, a binocular camera calibration module, a fully mechanized mining face SLAM module and a coal mine map maintenance and unmanned vehicle path planning module; the anti-explosion box is welded on a chassis of the unmanned vehicle, the binocular camera is mounted on the roof of the unmanned vehicle, and the top end of the binocular camera is provided with a protection frame; the sensor module comprises an inertial sensor;
the inertial sensor acquires the motion information of the unmanned vehicle;
the binocular camera acquires image information of the unmanned vehicle;
the coal mine environment image processing module establishes a mapping relation between image information acquired by the binocular camera and an actual calibration environment in a mine;
the fully-mechanized mining face SLAM module positions the position of the unmanned vehicle in the mine according to image information captured by the binocular camera and motion information acquired by the inertial sensor, and an initial coal mine environment map is created;
and after the position positioning information of the unmanned vehicle in the mine and the initial coal mine environment map information obtained by the fully mechanized mining face SLAM module are continuously updated by the mine map maintenance and unmanned vehicle path planning module, the optimal path of the unmanned vehicle to other positions in the mine is determined by the unmanned vehicle control system and a path planning algorithm adopting a rolling window.
The method for realizing autonomous patrol of the fully mechanized mining face patrol unmanned vehicle for the coal mine based on the SLAM technology comprises the following steps:
the first step is as follows: calibrating internal parameters of a binocular camera, initializing a fully mechanized mining face SLAM module, and obtaining an initial coal mine environment map;
the second step is that: when the unmanned vehicle patrols the mine, the binocular camera matches the image information acquired at the current position with the initial coal mine environment map for feature points, and if the matching is successful, the third step is carried out; if the matching is unsuccessful, the control system of the unmanned vehicle considers that the positioning quality of the current fully-mechanized mining face inspection unmanned vehicle is poor, local repositioning work of the fully-mechanized mining face inspection unmanned vehicle is triggered, namely image information acquired by the binocular camera is added into the initial coal mine environment map in the first step, and then feature point matching is carried out again until the matching is successful;
the third step: if the local map feature points are successfully matched, the unmanned vehicle adopts an ICP algorithm based on a least square method to establish a fully-mechanized mining face patrol unmanned vehicle motion model, an SVD decomposition method is used for solving the motion model, and the pose of the fully-mechanized mining face unmanned vehicle is solved by combining a nonlinear optimization mode; meanwhile, continuously adding environment map points of new positions in the patrol process of the unmanned vehicle to form a current coal mine environment map, and continuously updating the current coal mine environment map;
the fourth step: and after the inspection task corresponding to the current mine position by the unmanned vehicle is finished, determining the optimal path of the unmanned vehicle to the next target position in the mine through the unmanned vehicle control system and a path planning algorithm adopting a rolling window, and enabling the unmanned vehicle to reach the next target position by the optimal path to continue inspection work.
The unmanned vehicle can be used for autonomously positioning and constructing the environment at a plurality of positions of a coal mine fully mechanized mining face, a coal mine hydraulic support, a roadway and the like based on the visual SLAM technology, and meanwhile, the unmanned vehicle can be used for patrol monitoring and disaster early warning of the whole mine, so that workers can conveniently master the actual change condition of the fully mechanized mining face and the geographical position and environmental change below the whole mine, and can perform early warning and other work on the disaster, and the labor and the financial resources are saved to a great extent.
Detailed Description
In the following, the fully mechanized mining face patrol unmanned vehicle of the present application will be specifically described with reference to the accompanying drawings, which specifically include the following descriptions:
the schematic diagram of the model of the specific body of the fully mechanized face patrol unmanned vehicle is shown in figure 1. Including two mesh cameras 2, protection frame 1, chassis 4, explosion-proof box 3, the welding of explosion-proof box 3 is on synthesizing face inspection unmanned vehicles chassis 4, two mesh cameras 2 are installed on the roof, and two mesh cameras 2 top installation protection frames 1 prevent that coal cinder or other objects from rolling down pounding two mesh cameras.
The unmanned vehicle customized explosion-proof box 3 is formed by die-casting an aluminum alloy shell and spraying high-voltage static on the surface.
As shown in fig. 2, the fully mechanized mining face patrol unmanned vehicle is also provided with an intelligent module and an inertial sensor; the intelligent module comprises a coal mine environment image processing module, a binocular camera calibration module, a fully mechanized mining face SLAM module and a mine map maintenance and unmanned vehicle path planning module.
The unmanned vehicle control system consists of the English WEIDA TX2 development edition and an ROS operating system. The trolley employs an ROS operating system to enable communication between different hardware. The ROS system creates an ROS topic on the england TX2 development edition using an official development suite, and writes a control algorithm of the fully mechanized face patrol unmanned vehicle, a fully mechanized face SLAM algorithm, a path planning algorithm of a rolling window, and the like into the development board in advance.
As shown in FIG. 3, the technical problems of the fully mechanized mining face inspection unmanned vehicle based on SLAM technology of the invention, such as autonomous positioning, navigation, mine fully mechanized mining face inspection and the like, are solved through the following steps.
Firstly, calibrating internal parameters of a binocular camera carried by the unmanned vehicle, initializing a fully mechanized mining face SLAM module and acquiring an initial coal mine environment map due to the particularity of the working environment of the fully mechanized mining face patrol unmanned vehicle.
The specific steps of internal parameter calibration are as follows:
step 1) aligning a binocular camera to a fully mechanized mining face environment template to be calibrated;
step 2) starting a binocular camera calibration module, photographing the environment template of the fully mechanized mining face to be calibrated from at least two directions on the premise of ensuring that the environment template of the fully mechanized mining face to be calibrated of the unmanned vehicle is completely positioned in the field of view of the binocular camera image carried by the unmanned vehicle, and acquiring a calibration template image;
and 3) establishing a corresponding relation between the characteristic points in the acquired calibration module image and the characteristic points of the actual calibration environment, automatically calculating an internal parameter matrix optimized by a camera of the fully-mechanized face inspection unmanned vehicle by adopting a Zhang Zhengyou plane calibration algorithm, and storing the internal parameter matrix in a binocular camera calibration module.
Secondly, when the unmanned vehicle patrols the mine, matching the coal mine environment image information acquired at the current position with the initial coal mine environment map by using feature points, and if the matching is successful, performing a third step; if the matching is unsuccessful, the control system of the unmanned vehicle can determine that the position positioning quality of the current fully-mechanized mining face inspection unmanned vehicle is poor, and then local repositioning work of the fully-mechanized mining face inspection unmanned vehicle can be triggered.
The local relocation work specifically comprises the following steps:
step 1): adding local image information acquired by a binocular camera into an initial coal mine environment map established by the SLAM module;
step 2): and carrying out feature point matching on the acquired image information and a new environment map corresponding to the current working position of the unmanned vehicle by using a binocular camera.
The third step: if the feature points are successfully matched, the unmanned vehicle adopts an ICP algorithm based on a least square method to establish a fully-mechanized mining face inspection unmanned vehicle motion model, an SVD decomposition method is used for solving the motion model, and a nonlinear optimization mode is combined to solve the current pose of the fully-mechanized mining face unmanned vehicle; meanwhile, automatically updating the SLAM module of the fully mechanized mining face according to the matching result; and updating the map feature points of the fully mechanized mining face patrol unmanned vehicles at the current working positions, adding the updated map feature points into an initial coal mine environment map, and updating the map corresponding to the positions of the current unmanned vehicles in the coal mine.
The method adopts a Minimum Mean Square Error (MMSE) algorithm of a grey level histogram based on the main direction of the coal mine environment to match the characteristic points.
And fourthly, after the inspection task corresponding to the current mine position by the unmanned vehicle is completed, determining the optimal path of the unmanned vehicle to the next target position in the mine through the unmanned vehicle control system and a path planning algorithm adopting a rolling window, and enabling the unmanned vehicle to reach the next target position by the optimal path to continue the inspection work.
The route planning algorithm of the rolling window comprises the following specific steps:
step 1) initializing all parameters of the unmanned vehicle, such as a starting point, a target point and a working environment, setting the size of a rolling window, and inspecting the moving step length of the unmanned vehicle at each time on the fully mechanized mining face;
step 2) continuously updating the environmental information of the current rolling window, and if a dynamic barrier exists in the window, predicting the motion state of the barrier at the next moment according to a linear prediction model of the dynamic barrier; if no dynamic barrier exists, the mine unmanned vehicle plans a local optimal path moving towards the mine unmanned vehicle according to the local optimal sub-target point generated in the window;
step 3) according to the collision prediction, making a corresponding obstacle avoidance strategy for the fully mechanized mining face inspection unmanned vehicle;
step 4), the fully-mechanized mining face inspection unmanned vehicle stops for a period of time or changes the speed to find out a path which does not collide with the obstacle according to a corresponding obstacle avoidance strategy;
step 5) moving to the planned path according to the set steps, wherein the step length cannot be larger than the radius of the window;
and 6) if the fully mechanized mining face patrols the unmanned vehicle to reach the target position, finishing the algorithm, and the method is described in detail by combining a specific mode.
While preferred embodiments of the present invention have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. Therefore, it is intended that the appended claims be interpreted as including preferred embodiments and all such alterations and modifications as fall within the scope of the invention.
It will be apparent to those skilled in the art that various changes and modifications may be made in the present invention without departing from the spirit and scope of the invention. Thus, if such modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include such modifications and variations.