Research On Gait Planning And Walking Control Of Quadruped Crawling Robot On A Slope

The quadruped crawling robot will pass through the complex terrain when performing tasks on a slope.In order to ensure its passability on the slope,it is necessary to improve the control accuracy of the foot end,control the foot end force and identify the terrain effectively.Therefore,this paper studies the quadruped crawling robot from the aspects of multi gait design,coordination control of the leg joints,compliance control of the foot end force and slope environment identification.The main contents are as follows:First of all,the kinematics model of the quadruped crawling robot under the flat slope is established.Aiming at the flat,complex and convex terrain,the walking gait,obstacle avoidance gait and climbing gait are designed.In view of the fact that the gait switching requires the cooperation of the joint motor,the mapping relationship of the model between multi gait and joint motor is explored,and the temporal prediction control of the joint motor is proposed to achieve the soft switch of the multi gait.Secondly,in order to extend the movement time of the quadruped crawling robot on a slope,the leg energy consumption model is established,and the energy consumption of the robot is optimized according to the movement space of the leg on the slope.Aiming at the problem of poor motion coordination caused by leg joint coupling,a neural network model reference decoupling control based on interference observation is designed.The position simulation of the hip and knee joint under MATLAB verifies that the decoupling control has higher accuracy.Thirdly,when the legs of the quadruped crawling robot are switched from the swing phase to the support phase,the contact force between the foot end and the ground is increased due to the existence of slope,which reduces the stability of the robot's movement,so a force/position segmented switching control is proposed.When the leg is in the swing phase,the variable speed PID positionalgorithm is used to realize the precise positioning.When the foot end contacts the ground,the impedance control based on the position is used to achieve the compliance control of the force during the support stage.In the force/position switching,considering the influence of the friction between the foot and the slope,a fuzzy multi-model switching algorithm based on friction compensation is used to achieve smooth switching between force and position.The force and position simulation under MATLAB verifies that the proposed algorithm has better dynamic stability.At last,in order to improve the ability of the quadruped crawling robot to recognize the terrain in real time,the monocular camera is used to acquire the front environment that ensuring the robot's passability on a slope.Considering that the recognition speed of the slope environment of the quadruped crawling robot is fast and needs high positioning accuracy,the ORB feature points are extracted by setting the adaptive threshold value and saved by the quadtree.In the matching stage,the FLANN algorithm is used for feature matching to realize the real-time identification of the slope terrain,and according to the slope terrain,the forward,obstacle avoidance and climbing strategies are made.The appropriate landing point is obtained through the continuous interaction with the slope environment by using the search-based reinforcement learning,and the prediction model based on the dynamic model is used to effectively track the target trajectory of the leg joints to ensure the accuracy of the position of the landing point.The motion simulation in ROS(Robot Operating System)environment verifies that the quadruped crawling robot can pass the slope in real time and stably.