iDRM & Improved MaxiMin NSGA-II-based Motion Planning for a Humanoid Mobile Manipulator System

Abstract

Autonomic mobile base (MB)'s location & upper-body's configuration design and control are two of the fundamental issues to realize robotic autonomy. Besides, the human-like behaviors capacity is essential for human-robot interaction scenarios. Thus, to achieve human-like autonomy, this paper proposes a motion planning and tracking strategy for a redundant dual-arm humanoid mobile manipulator system. Specifically, the inverse dynamic reachability map (iDRM), the improved MaxiMin NSGA-II and the direct-connect bidirectional RRT & gradient descent algorithms are employed. Firstly, iRM is constructed and used to design potential MB's location area. Then, the improved MaxiMin NSGA-II algorithm is used to determine the desired end-pose based on the selected MB's location area. After the desired end-pose is designed, the direct-connect bidirectional RRT & gradient descent algorithm will be used to generate the approaching trajectory from the initial pose to the desired end-pose. Finally, the RBFNN controller is used to track the planned trajectory. On the other hand, if there are new obstacles, iDRM will be obtained by updating iRM. iRM and iDRM accelerate the desired end-pose design by narrowing MB's searching area. Moreover, to achieve human-like motions five criteria including the end-effectors' (EE) displacement with respect to MB are proposed. Several motion planning simulations are realized that validate the proposed strategy.