Remote Supervision of Autonomous Humanoid Robots for Complex Disaster Recovery Tasks Stefan Kohlbrecher, TU Darmstadt Joint work with Alberto Romay, Alexander Stumpf, Oskar von Stryk Simulation, Systems Optimization and Robotics Group, Technische Universität Darmstadt, Germany Felipe Bacim, Doug A. Bowman Center for Human-Computer Interaction, Virginia Tech, United States David C. Conner TORC Robotics, United States
Complex Disaster Recovery DARPA Robotics Challenge (DRC) Limited capability in existing disaster response robots Fukushima nuclear plant disaster Provide human-like capabilities Beyond inspection Ability to use tools made for humans Intuitive understanding for operators Task level autonomy Restricted communication 2
The C in DRC Uneven terrain, stairs and ladders Motions with multiple contacts (e.g. getting into and out of a vehicle) (Loco-)Motion 3
The C in DRC Perception (Loco-)Motion Environment for locomotion Objects for manipulation Ability to acquire new objects and their potential purposes on the fly Robustness to different lightning conditions 4
The C in DRC Perception Manipulation (Loco-)Motion Many different tools, only few exactly known in advance Acquiring new manipulation modes Ability to coordinate manipulation, locomotion & active perception 5
The C in DRC Perception Manipulation Efficient interaction with (Loco-)Motion Human Supervisor Matching human and robot abilities best Appropriate levels of human-robot-interaction for highly diverse tasks Distribution between work tasks robot onboard and offboard (OCS) 6
The C in DRC Perception Manipulation Efficient interaction with (Loco-)Motion Human Supervisor Limited Wireless Communication bandwidth, delays, losses 7
Human-Robot Team Leverage synergies between operator and onboard system Challenges: Provide operator with situational awareness Translating operator intent to robot action 8
Translating Operator Intent - Object Templates Sense Kohlbrecher et al Human-Robot Teaming for Rescue Missions: Team ViGIR s Approach to the 2013 DARPA Robotics Challenge Trials Journal of Field Robotics, 2014 Romay et al. Template-Based Manipulation in Unstructured Environments for Supervised Semi-Autonomous Humanoid Robots, IEEE Humanoids 2014 Stumpf et al. Supervised Footstep Planning for Humanoid Robots in Rough Terrain Tasks Using A Black Box Walking Controller, IEEE Humanoids 2014 9
Translating Operator Intent - Object Templates Sense Plan Kohlbrecher et al Human-Robot Teaming for Rescue Missions: Team ViGIR s Approach to the 2013 DARPA Robotics Challenge Trials Journal of Field Robotics, 2014 Romay et al. Template-Based Manipulation in Unstructured Environments for Supervised Semi-Autonomous Humanoid Robots, IEEE Humanoids 2014 Stumpf et al. Supervised Footstep Planning for Humanoid Robots in Rough Terrain Tasks Using A Black Box Walking Controller, IEEE Humanoids 2014 10
Translating Operator Intent - Object Templates Sense Plan Kohlbrecher et al Human-Robot Teaming for Rescue Missions: Team ViGIR s Approach to the 2013 DARPA Robotics Challenge Trials Journal of Field Robotics, 2014 Romay et al. Template-Based Manipulation in Unstructured Environments for Supervised Semi-Autonomous Humanoid Robots, IEEE Humanoids 2014 Stumpf et al. Supervised Footstep Planning for Humanoid Robots in Rough Terrain Tasks Using A Black Box Walking Controller, IEEE Humanoids 2014 11
Translating Operator Intent - Object Templates Sense Plan Kohlbrecher et al Human-Robot Teaming for Rescue Missions: Team ViGIR s Approach to the 2013 DARPA Robotics Challenge Trials Journal of Field Robotics, 2014 Romay et al. Template-Based Manipulation in Unstructured Environments for Supervised Semi-Autonomous Humanoid Robots, IEEE Humanoids 2014 Stumpf et al. Supervised Footstep Planning for Humanoid Robots in Rough Terrain Tasks Using A Black Box Walking Controller, IEEE Humanoids 2014 12
Translating Operator Intent - Object Templates Sense Plan Walk Grasp Use Kohlbrecher et.al Human-Robot Teaming for Rescue Missions: Team ViGIR s Approach to the 2013 DARPA Robotics Challenge Trials Journal of Field Robotics, 2014 Romay et al. Template-Based Manipulation in Unstructured Environments for Supervised Semi-Autonomous Humanoid Robots, IEEE Humanoids 2014 Stumpf et al. Supervised Footstep Planning for Humanoid Robots in Rough Terrain Tasks Using A Black Box Walking Controller, IEEE Humanoids 2014 13
DRC Hose Task 14
DRC Hose Task https://www.youtube.com/watch?v=i8pb6gpvleo 15
DRC Hose Task 16
Summary Brief intro to the DRC Overview of Team ViGIR s Human-Robot teaming approach Key challenges: How to match robot and human capabilities best Situational Awareness Approaches for interaction Outlook: DRC Finals in June 2015 Team ViGIR (Altas robot) Team Hector (TU Darmstadt, THORMANG robot) 17