Research On Path Planning And Control Algorithm Of Wheel-legged Planet Surface Exploration Robot

With the deepening of deep space exploration missions,the future planet surface roving exploration will develop towards large range,long period and multiple targets.The exploration area will be expanded from the relatively flat area at present to the rugged complex area,which puts forward higher requirements for the roving probe’s speed and terrain adaptability.In view of the requirements of future planet surface exploration missions,this paper developed a wheel-legged planet surface exploration robot.Considering the high degree of unknown of planet surface,variable configuration of the wheel-legged robot,and high efficiency requirement of control algorithm,research on hierarchical path planning algorithm and time-efficient wheeled motion and legged motion control algorithm for wheel-legged robot is carried out.The feasibility and superiority of the proposed algorithms are proved by theoretical analysis,simulation test and prototype experiment.(1)A wheel-legged planet surface exploration robot with high speed and strong terrain adaptability is developed and introduced for future planet surface roving exploration missions.On that basis,the kinematic and dynamics of the robot is analyzed and modeled from the top to the bottom.The robot simulation model is built based on Matlab/Simulink,which provides necessary basis for the verification and analysis of path planning and control algorithms.(2)Aiming at the problem of complexity caused by the variable configuration of the wheel-legged robot,this paper proposes a path hierarchical planning algorithm considering the ablility of crossing obstacle of torso and wheel.The obstacles are divided into complete obstacles and incomplete obstacles according to the height,and virtual obstacles of the torso were determined according to the relative position between the horizontal projection of the workspace of the single wheel-leg end,torso and the obstacle area.The torso passability map and wheel passability map are constructed,respectively.The Theta* algorithm is used to plan the original torso path,and the timed elastic band algorithm is used to smooth the original torso path.The wheel path is planned based on ergodic search,and the hierarchical path planning of the wheel-legged robot is realized.The effectiveness of the proposed algorithm is proved by the tests in complete obstacle map,incomplete obstacle map and mixed obstacle map.Compared with traditional algorithms,the overexpansion of obstacle is avoided and the path length and planning time are reduced up to 38.3% and 83.0%,respectively.The advantages of the proposed method in terrain adaptability,path length and planning time are proved.(3)Aiming at the problem of high degree of unknown of planet surface and poor computing ability of the roving robot control system,this paper proposes a timeefficient wheeled motion control algorithm based on velocity planning and torso state estimation according to the joint encoder and inertial measurement unit,which is divided into the robot velocity planning layer and the single wheel-leg impedance control layer.The desired velocity of the torso was planned by the robot velocity planning layer according to the torso position and attitude deviation,and then the desired velocity of the single wheel-leg is planned according to the rigid body velocity transformation formula.The operation of matrix inverse or optimization problem solving in the force distribution process of virtual model control algorithm is avoided,and the collaborative velocity planning of the whole wheel-legged robot is realized.The impedance control layer planned the expected force and moment of the end of single wheel-leg according to the velocity error of the end of single wheel-leg and its integral.The force Jacobian matrix of the single wheel-leg is utilized to obtain to the joint torque to realize the compliance control of the robot.On that basis,the time complexity of the algorithm is analyzed from the aspects of basic operation times and progressive time complexity.By comparing with the improved virtual model control algorithm,the superiority of the proposed method in time complexity is proved.Finally,through the simulation and experiment under the conditions of slope terrain and unilateral bridge terrain,it is proved that the proposed algorithm can ensure the stable contact between wheel and the ground under the fluctuated terrain,while keeping the posture of the torso relatively stable.In the simulation of slope terrain,the peak value of body posture is only 17.1% of the angle of slope and the adaptability to terrain is realized effectively.(4)In view of the more stringent computational requirement of legged motion control algorithm,this paper proposes a time-efficient legged motion control algorithm based on function allocation and velocity planning.The leg motion sequence is planned on the basis of static stability analysis of the wheel-legged robot.Based on trigonometric function,the center of gravity adjustment trajectory satisfying the constraints of position,velocity and acceleration and the rectangular trajectory of the single wheel-leg end are planned.Then,the functions of the six active degrees of freedom of the single wheel-leg are allocated.On that basis,a feedforward item which allocates the contact force is added to realize overall cooperative control of support phase of legged motion,and the change of wheel-ground contact force is reduced at the moment robot lifts its leg.The tracking control of the end of the single wheel-leg in the swing phase is realized based on the impedance control algorithm.The time complexity of the algorithm is analyzed from the point of view of basic operation times and progressive time complexity.The result proves that the legged motion control algorithm has a lower time complexity than the wheeled motion control algorithm.Simulation and experiment are carried out to adjust the horizontal position and the vertical position of the torso,the contact force distribution,and static legged motion which proves the effectiveness of the proposed algorithm.Combined with the feedforward item of force distribution,the pitch angle and roll angle of torso at the moment of lifting leg are reduced up to 22.2% and 30.8%,respectively and the stability of torso is improved effectively.