3D Motion Planning for Flexible Needle Insertion Based on Rapidly-Exploring Random Trees with Environment-Adaptive Sampling and Central Angle Control

The utilization of flexible needles in puncture surgery can facilitate obstacle avoidance and solve the problem of uncertain deformation of needle body and tissue. Motion planning for the bevel-tip flexible needle is crucial but challenging due to the nonholonomic constraints of flexible needle and the presence of obstacles. Rapidly-exploring random trees (RRTs) method is one very popular technology for handling nonholonomic constraints and high degrees of freedom. However, the feasible paths generated by the previous RRTs-based algorithms may be infeasible in clinical practice due to their large detour problem and the convergence problem. Meanwhile, the pseudo feasible paths in candidate path set produced through re-running will seriously influence the computational efficiency and stability of these methods in the process of obtaining one sub-optimal or optimal path through optimization. Thus, we have developed one improved RRTs-based approach by combining the reachability-guided strategy, the greedy heuristic strategy and the central angle control method to force the paths generated by our algorithm to be excellent enough and be clinically executable. Besides, we have adopted the environment-adaptive sampling strategy to improve the computational efficiency and stability of our algorithm. The simulations and experiments demonstrate that our algorithm can perform the more stable and faster motion planning than the compared algorithms.