Quantcast

Collaborative Autonomy between High-level Behaviors and Human Operators for Remote Manipulation Tasks using Different Humanoid Robots

Research paper by Alberto Romay, Stefan Kohlbrecher, Alexander Stumpf, Oskar von Stryk, Spyros Maniatopoulos, Hadas Kress‐Gazit, Philipp Schillinger, David C. Conner

Indexed on: 10 Oct '16Published on: 08 Sep '16Published in: Journal of Field Robotics



Abstract

Team ViGIR and Team Hector participated in the DARPA Robotics Challenge (DRC) Finals, held June 2015 in Pomona, California, along with 21 other teams from around the world. Both teams competed using the same high-level software, in conjunction with independently developed low-level software specific to their humanoid robots. On the basis of previous work on operator-centric manipulation control at the level of affordances, we developed an approach that allows one or more human operators to share control authority with a high-level behavior controller. This collaborative autonomy decreases the completion time of manipulation tasks, increases the reliability of the human-robot team, and allows the operators to adjust the robotic system's autonomy on-the-fly. This article discusses the technical challenges we faced and overcame during our efforts to allow the human operators to interact with the robotic system at a higher level of abstraction and share control authority with it. We introduce and evaluate the proposed approach in the context of our two teams' participation in the DRC Finals. We also present additional, systematic experiments conducted in the lab afterward. Finally, we present a discussion about the lessons learned while transitioning between operator-centered manipulation control and behavior-centered manipulation control during competition.