Squat motion of a bipedal robot using real‐time kinematic prediction and whole‐body control

Squatting is a basic movement of bipedal robots, which essential in robotic actions like jumping or picking up objects. Due to the intrinsic complex dynamics perfect squatting motion requires high-performance planning and control algorithms. The standard academic solution combines model predictive (MPC) with whole-body (WBC), usually computationally expensive difficult implement on practical robots limited computing resources. real-time kinematic prediction (RKP) method proposed, considers upcoming reference trajectories it quadratic programming (QP)-based WBC. Since WBC handles full robot various constraints, RKP only needs adopt linear kinematics robot's task space softly constrain desired accelerations. Then, computational cost derived closed-form greatly reduced. verified simulation heavy-loaded robot. makes rapid large-amplitude motions, require close-to-limit torque outputs. Compared conventional QP-based method, proposed exhibits high adaptability rough planning, implies much less user interference planning. Furthermore, MPC, can prepare for motions advance but computation time.