Research On Autonomous Obstacle Negotiation Of Wheel-legged Mobile Robot Based On Obstacle Detection

The robots have been used more and more in the production and life of human beings.Mobile robots with obstacle negotiation capabilities are mostly used in rescue and disaster relief,scientific exploration and military operations due to their strong environmental adaptability.This research takes the wheel-legged mobile robot used in the field combat environment as the research object to carry out the research on the autonomous obstacle negotiation function.The robot in this research relies on the structural characteristics of its suspension and adjustable wheelbase,by changing its own shape to achieve the purpose of overcoming obstacles.The obstacle negotiation is mainly for three types of scenarios: positive obstacles,negative obstacles,and trench obstacles.The autonomy of obstacle negotiation is based on the detection of obstacles,which are based on real-time maps.Firstly,a stepwise approach is used to plan the robot's obstacle negotiation actions for the above three types of obstacles.And then,the centroid kinematics model is established to obtain the robot's centroid formula,and the stability margin model is established based on the centroid projection method to obtain its stability margin angle formula.By selecting the key posture during obstacle negotiation,the obstacle negotiation ability and stability are analyzed,and the control variables in the obstacle negotiation are determined.Secondly,the detection of the above three types of obstacles is realized based on the elevation map constructed in real time.Based on the Grid Map Library,an elevation map algorithm that can follow robot movement and display is developed.Before the detection,the ground segmentation algorithm based on plane fitting is used to separate the ground from the obstacle in the map.For the detection of positive obstacles,a algorithm based on the smallest rectangular bounding box is proposed.The algorithm will calculate the box closest to the shape of the obstacle.As for the negative obstacles and trench obstacles,they are initially positioned according to the lack of data in the elevation map,and then the specific types of obstacles are determined through further detection.The detection algorithm calculates the position and size of the obstacle in detail.At the same time,two judgment indexes are proposed,and the judgment standard of whether the obstacle has the obstacle negotiation condition is given by combining the judgment index and the size of the obstacle.Thirdly,it realizes autonomous obstacle negotiation control and builds a system framework of autonomous obstacle negotiation.The core functional modules of the autonomous obstacle negotiation system include the above-mentioned obstacle negotiation planning and analysis,mapping and obstacle detection algorithms,and other auxiliary function modules and autonomous obstacle negotiation control methods are responsible for providing basic functions and connecting the entire system in series,,which mainly includes: realizing the localization function of the robot using laser SLAM,realizing the posture adjustment of the robot before negotiation using the ROS's navigation module,establishing the real-time perception of the spatial position relationship between the robot and the obstacle through a feature extraction method based on two-dimensional laser scanning lines,and thus realizes the control of the obstacle negotiation process.Finally,the joint simulation experiment of Adams and Simulink verifies the feasibility of the robot obstacle negotiation planning.The simulation experiments under ROS and Gazebo prove that the obstacle detection method has good detection ability and detection accuracy,as well as the robot autonomous obstacle negotiation system has good performance.