Maximizing Task-Oriented Manipulability of Redundant Robots

Abstract

Approaching human-level performance in robotics allows to perform task of increasing complexity. This is used in assistive and rehabilitation robotics to facilitate the daily life of physically disabled people. However, assistive robotic systems such as EDAN are subject to kinematic and physical limitations regarding joint alignment, velocity and operating range. Those limits can prevent reaching a task’s goal if they are not compensated for properly. Based on the standard manipulability ellipsoid introduced by Yoshikawa in [1], a new deformed manipulability volume has been developed including kinematic and physical limits for the first time in a real-time capable approach. Optimizing for the distance from the center to the surface (called transmittion ratio or directional manipulability) allows to avoid singularities and joint limits that are obstacles to moving in this specific direction of interest. Other singularities or joint limits can be approached by intention in order to exploit the full capabilities of the robotic system. This new theory has been tested extensively in simulation and on the real robotic system, illustrating theoretical superiority compared to state-of-the-art methods. As a result, it is part of a recent patent claim by Maged Iskandar, Jörn Vogel and Frederic Sauer.