A Growth-State and Branch-Tip Based Adaptive Bi-RRT for Cartesian Path Planning of Redundant Manipulators

Cartesian-space path planning for redundant manipulators in complex environments often suffers from low search efficiency, poor path quality, and stagnation in narrow passages. This paper proposes an adaptive bidirectional rapidly-exploring random tree (Bi-RRT) algorithm based on growth-state awareness and branch-tip direct connection. An adaptive goal sampling strategy and dynamic step-size adjustment are introduced to balance exploration and goal guidance. A growth-state switching mechanism enables efficient escape from constrained regions, while a branch-tip direct connection strategy reduces connectivity checking cost and accelerates convergence. Simulation results on a Franka Emika Panda manipulator demonstrate that the proposed method significantly improves planning efficiency and generates shorter collision-free paths in various environments.