Body load position estimation and cooperative control method based on robot body perception

Abstract

The application provides a body load position estimation and cooperative control method based on robot body perception, and relates to the technical field of biped robot control. The method uses robot joint sensors and on-board IMU, removes leg weight and inertia interference through dynamic preprocessing, extracts load external torque by using a momentum observer, maps and calculates torque balance through a Jacobian matrix, and estimates the mass, position and speed of a non-fixed dynamic load in real time. The estimation distortion caused by body posture tilt is eliminated by projection correction through a rotation matrix, and smooth output is realized in combination with Alpha-Beta filtering. The control link maps the load movement into body posture instructions by using a backstepping method, outputs joint torque through inverse kinematics, a second-order tracking differentiator and a second-order Super-Twisting sliding mode controller, and realizes robot self-balancing and load cooperative control. The application can effectively solve the problems of traditional schemes, such as dependence on external sensors, leg dynamics coupling interference, dynamic load estimation lag, posture tilt positioning distortion and the like.

Images