Weighted Shared-Autonomy with Assistance-to-Target and Collision Avoidance for Intelligent Assistive Robotics

Intelligent Assistive Robotics (IAR) has been recently introduced as a branch of Service Robotics developing semi-autonomous robots helping people with physical disability in daily-living activities. Literature often focuses on the development of assistive robots with a single semi-autonomous behavior, while the integration of multiple assistance is rarely considered. In this paper, we propose a novel shared-autonomy controller integrating the contribution of two semi-autonomous behavioral modules: an assistance-to-target module, adjusting user’s input to simplify the target reaching, and a collision avoidance module, moving the robot away from trajectories leading to possible collisions with obstacles. An arbitration function based on the risk of collision is introduced to prevent conflicts between the two behaviors. The proposed controller has been successfully evaluated both offline and online in a reach-to-grasp task with a simulated robotic manipulator. Results show that the proposed methods significantly reduced not only the time to complete the task with respect to the pure teleoperation or controllers including just one semi-autonomous behavior, but also the user’s workload controlling the manipulator with a wearable interface. The context-awareness employed by the IAR may increase the reliability of the human-robot interaction, pushing forward the use of this technology in complex environments to assist disabled people at home.