| Pre-exploration of post-disaster environment by mobile robot is a mainstream means of post-disaster relief in various countries.As a mobile robot with composite structure,wheel-track legged robot has outstanding motion performance and can efficiently complete environmental exploration tasks in complex post-disaster environment.Robot selfmodeling is the basis of motion control during the task execution in the highly unstructured post-disaster environment and the change of the environment causes the change of the robot body attitude.The attitude perception method is the cornerstone of the robot self-perception,and the attitude control method is the key to the stability of the robot.Based on the subject background of the national key research and development program"wheel-track legged autonomous vehicle development",this paper takes the wheel-track legged robot independently developed by the laboratory as the research object,and studies the motion modeling technology,attitude perception and prediction methods and attitude control methods of the wheel-track legged robot.The main research work is as follows:Firstly,the floating base idea is introduced to study the system modeling of wheel-track legged robot under complex rescue conditions.Taking the self-developing wheel legged robot as the research object,the motion constraints of the robot were analyzed from the movement characteristics of the fuselage,and the translation and attitude movements of the coupled wheel legged robot were decoupled.A floating base system was introduced to establish the pose mapping relationship between the end of the robot swinging arm and the Angle space of the robot joint.Meanwhile,the conversion relationship between the joint velocity and the operating space velocity,the joint torque and the generalized force was established.In the fixed coordinate system,the inverse dynamics model of the pendulum arm was established,and the dynamic system under the floating basis was projected into the constrained null space.The driving torque of the decoupling joint was decomposed by the constrained Jacobian matrix transposed QR,and the contact force was derived.Secondly,in order to obtain the real time attitude of the robot,the attitude perception and prediction method of the wheel-shoe legged robot are studied.The information processing method of each sensor and the initial calibration method of motion observation scale were studied,and the front-end processing of multi-sensor information fusion was completed.The sliding window model of robot motion observation was constructed,the residual error of robot motion observation was analyzed,and the nonlinear least squares problem of motion state optimization vector was studied.The nonlinear optimization fusion of multi-sensor information was completed,and the attitude perception efficiency of wheel-legged robot was effectively improved.Thirdly,the attitude control of wheel-shoe legged robots with asymmetric load was studied considering the asymmetric load.Based on the analysis of the influence of the swinging arm movement on the center of gravity migration of the wheel-legged robot,the influence of inertia on the robot stability was considered,and the motion attitude stability of the robot was studied by combining ZMP and FASM.The coupling part,the nonlinear part,the unmodeled part and the external interference of the system are unified.The attitude controller of the robot is designed,and the stability of the control system is analyzed.By analyzing the force on the end of the swing arm of the wheel-shoe legged robot without load,the center of gravity of the robot can be estimated in real time.Finally,the attitude perception and control methods of wheel-shoe legged robots are studied experimentally.Build the prototype experiment platform,design the experiment process and carry out the physical experiment.By comparing the experimental results,the correctness and effectiveness of the proposed algorithm are verified. |
PDF Full Text Request