Hierarchical Motion Planning Method for Space Redundant Manipulators with End-Task Constraint

This paper proposes a hierarchical motion planning method integrating goal-biased rapidly exploring random tree (GB-RRT) and sequential convex programming (SCP) for a class of space redundant manipulators (SRM) utilizing an end hand–eye camera to perform visual monitoring tasks. By analyzing the main influencing factors in an SRM system, the nonlinear optimal control problem model that incorporates kinematics, dynamics, obstacle avoidance, and line-of-sight constraints is first developed to describe the trajectory planning of SRM. Next, to warm-start the solution of the SCP program, a trajectory planning algorithm based on GB-RRT and cubic spline curves is designed at the kinematics level to rapidly plan collision-free and smooth reference trajectories. Then, all the constraints at the dynamics level are convexified with the help of the velocity damping method, quaternion differentiation, first-order Taylor expansion, and continuous linearization techniques. As a result, the optimal solution of the original planning problem can be obtained by solving the new SCP problem, in which the accuracy of the solution is guaranteed by applying trust region constraints to the key variables. The proposed hierarchical planning method not only avoids the poor scalability issue of the sampling-based motion planning method when considering complex constraints, but also provides high-quality initial reference trajectories for solving the SCP program. Finally, some numerical simulations are provided to verify the effectiveness and validity of the proposed method.