Sung-Hee Lee [KAIST, Korea]
Title: “Damage Minimization of Humanoid Robots by Falling on Targeted Body Segments”
Safety and robustness will become critical issues when humanoid robots start sharing human environments in the future. In physically interactive human environments, a catastrophic fall is a major threat to the safety and smooth operation of humanoid robots. It is, therefore, imperative that humanoid robots be equipped with a comprehensive fall management strategy. A common approach is to employ damage-resistant design and apply impact-absorbing material to robot limbs, such as the backpack and knee, that are particularly prone to fall related impacts. We select the backpack to be the most preferred body segment to experience an impact. We proceed to propose a control strategy that attempts to reorient the robot during the fall such that it impacts the ground with its backpack. We show that the robot can fall on the backpack even when it starts falling sideways. This is achieved by generating and redistributing angular momentum among the robot limbs through dynamic coupling. The planning and control algorithms for a fall are demonstrated in simulation.
Sung-Hee Lee is an Associate Professor with the Graduate School of Culture Technology at KAIST. He received the Ph.D. degree in Computer Science from University of California, Los Angeles, USA, in 2008, and the B.S. and the M.S. degree in Mechanical Engineering from Seoul National University, Korea, in 1996 and 2000, respectively. Prior to joining KAIST in 2013, he was an Assistant Professor at Gwangju Institute of Science and Technology from 2010 to 2013, and a Postdoctoral researcher at UCLA from 2008 to 2009, and at Honda Research Institute, CA, from 2009 to 2010. His research interests include physics-based computer graphics/animation and humanoid robotics, specifically in autonomous human characters and humanoid robots, biomechanical human modeling, and physics simulation.
Abderrahmane Kheddar [CNRS, France]
Title: “Reducing Damage in Humanoid Falls using Closed-Loop Posture Reshaping and Active Compliance.”
In this talk, I will discuss few ideas on how damage can be reduced when humanoid falls are unavoidable. We are working toward deploying humanoid technology in large-scale manufacturing yards, e.g. aircraft manufacturing (www.comanoid.eu); for such applications, falls should be assumed to be unavoidable. In these contexts, humanoid robots should prove that they know how to fall. Hence, we need to design an approach that is first plausible and not necessary “fancy” in terms of advanced skills. For example, rolling as exhibited in parkour or some martial art sport is not a plausible solution inside an airplane in the stage of assembly as the space is often confined. We have therefore chosen the direction to consider how to reduce damage when falls are unavoidable. Often, it is better just to fall instead of taking risks to maintain balance by trying to plan for an additional contact. Our approach is to build a controlled strategy around the ideas that: (i) there are few falling configurations that a robot can try to reach when possible, (ii) a configuration can be taken to increase the range of motions to absorb impact with the robot body, (iii) adapt the PD gains to comply with the impacts. I will show videos where the HRP-4 (a human size humanoid) is pushed frankly and discuss further research directions and lessons learned from preliminary experiments.
Abderrahmane Kheddar received the BS in Computer Science degree from the Institut National d’Informatique (ESI), Algiers, the MSc and Ph.D. degree in robotics, both from the University of Pierre et Marie Curie, Paris. He is presently Directeur de Recherche at CNRS and the Director of the CNRS-AIST Joint Robotic Laboratory (JRL), UMI3218/RL, Tsukuba, Japan. He is also leading the Interactive Digital Humans (IDH) team at CNRS-University of Montpellier LIRMM, France. His research interests include haptics, humanoids and recently thought-based control using brain machine interfaces. He is a founding member of the IEEE/RAS chapter on haptics (acting also as a senior advisor), the co-chair and founding member of the IEEE/RAS Technical committee on model-based optimization, Editor of the IEEE Transactions on Robotics and within the editorial board of some other related robotics journals; he is a founding member of the IEEE Transactions on Haptics and served in its editorial board during three years (2007-2010). He is an IEEE senior member and titular full member of the National Academy of Technology of France.
Sven Behnke [Univ of Bonn, Germany]
Title: “Instability Detection and Disturbance Rejection for Bipedal Walking.”
Bipedal walking is inherently unstable. Unmodeled aspects of the dynamics or disturbances easily lead to falls. We developed methods for detecting instabilities and rejecting disturbances in the context of humanoid soccer where falls are part of the game. The robot dynamics is modeled as a linear inverted pendulum. We adapt timing and placement of the next step based on the estimated balancing state. Our approach has low perceptual and computational requirements and was demonstrated with our soccer robot Dynaped. We also investigated the learning of model parameters. Experiments with our robot Copedo showed that effective capture steps can be learned from the first fall.
More information: http://www.nimbro.net/Humanoid
Sven Behnke is full professor for Autonomous Intelligent Systems at the University of Bonn. He received his MS degree in Computer Science (Dipl.-Inform.) in 1997 from Martin-Luther-Universität Halle-Wittenberg and obtained a PhD in Computer Science (Dr. rer. nat.) in 2002 from Freie Universität Berlin. He spent the year 2003 as postdoctoral researcher at the International Computer Science Institute, Berkeley, CA. From 2004 to 2008, he headed the Humanoid Robots Group at Albert-Ludwigs-Universität Freiburg. His research interests include cognitive robotics, computer vision, and machine learning. His team NimbRo won many international robot competitions, including five times in a row the RoboCup Humanoid TeenSize Soccer tournament and three times in a row the RoboCup@Home League for domestic service robots.
Karen Liu [Georgia Tech, USA]
Title: “Contact Planning for Minimizing Damage of Humanoid Falls.”
A humanoid in an interactive environment is often exposed to the risk of falling due to unexpected contacts or perturbations. A fall can potentially cause detrimental damage to the robot and enormous cost to repair. To reduce the likelihood of damaging robots during online operations, researchers and engineers often cover the robot exterior with soft guards to absorb the impact of falls. Though practical, these extra parts can potentially limit the range of motion or change the contact behaviors.
Alternatively, the robot can learn how to fall safely like humans do. We introduce a new planning algorithm to minimize the damage of humanoid falls by utilizing multiple contact points. Given an unstable initial state of the robot, our approach plans for the optimal sequence of contact points such that the initial momentum is dissipated with minimal impacts on the robot. Instead of switching among a collection of individual control strategies, we propose a general algorithm which plans for appropriate responses to a wide variety of falls, from a single step to recover a gentle nudge, to a rolling motion to break a high-speed fall. Our algorithm transforms the falling problem into a sequence of inverted pendulum problems and use dynamic programming to solve the optimization efficiently.
C. Karen Liu is an associate professor in School of Interactive Computing at Georgia Tech. She received her Ph.D. degree in Computer Science from the University of Washington in 2005. Liu’s research interests are in computer graphics, animation, and robotics. She develops computational models of human and animal motion and applies them to build tools to facilitate scientists, engineers, and artists to model, simulate, and design natural motion. Liu received a National Science Foundation CAREER Award, an Alfred P. Sloan Fellowship, and was named Young Innovators Under 35 by Technology Review. In 2012, Liu received the ACM SIGGRAPH Significant New Researcher Award for her contribution in the field of computer graphics.
Title: Online Detection of Balance Status of Legged Robots towards Versatile Whole-body Locomotion
During the past few years, the human civilisation has witnessed the Dawn of Humanoid Robots: a league of humanoids, including Walk-Man from IIT, Atlas from Boston Dynamics, Valkyrie from NASA, DRC-Hubo from KAIST, HRP robots, and S-One from SCHAFT, just to name a few.
Supposedly their graceful movements are yet to be appreciated, if the loss of equilibrium is not predictable beforehand, especially followed by headlong falls. Lest falling occurs to these beautifully engineered human artifacts, we ought to benchmark a set of different criteria such as Energy based Fall Prediction (EFP), Capture Point (CP) and Foot Rotation Indicator (FRI), to understand what physical quantities indicate real-time balance status of legged robots, and therefore, prevent tumble if possible. Also, based on our study, we will formulate how the very same principle can be applied to controlling dynamic walking by changing foot placement, as well as reflex behaviours using upper-limbs for fall recovery. Lastly, we are delighted to share some know-hows of hardware design that empowers fall resistance, in case all these control preventions fail.
Zhibin LI is currently a senior postdoctoral researcher in the Department of Advanced Robotics, Italian Institute of Technology (IIT). He graduated from the Harbin Institute of Technology (HIT) with the excellent graduate award in 2007, and obtained the joint PhD degree from University of Genova and IIT in 2012. He is currently nominated for the award of “Young Innovators 2016” organized by MIT Technology Review. He has general research interest in the control of dynamic systems, and most of his works are involved in the dynamic legged locomotion: bipedal walking, balance recovery, stabilization, and torque/impedance control of humanoid robots.
Title:Fall Prevention, Damage Reduction, and Novel Locomotion Style to Avoid Falling for Legged Robot
Abstract:We will introduce our past and ongoing research related to falling of legged robots. We have proposed a falling avoidance mechanism which can expand the support polygon rapidly with the expansion of four arms mounted on the foot. However, it is not realistic to assume that falling can be prevented entirely by using the proposed foot mechanism. Therefore, we also research on a flexible landing controller to reduce serious damage on the robot. We are focusing on the forward fall issue and investigating the effectiveness of active impedance for forward falls. Lastly, we will present a novel locomotion style for legged robot which can inherently avoid falling in an unstructured environment.
Kenji Hashimoto is an Assistant Professor of the Waseda Institute for Advanced Study, Waseda University. He received the B.E. and M.E. degrees in Mechanical Engineering from Waseda University, in 2004 and 2006, respectively, and he received the Ph.D. degree in Integrative Bioscience and Biomedical Engineering from Waseda University, in 2009. He received the IEEE Robotics and Automation Society Japan Chapter Young Award in 2006, the JSME Fellow Award for Outstanding Young Engineers in 2008, and the RSJ Young Investigation Excellence Award in 2015. His research interests include walking systems, biped robots, and humanoid robots.
Gan Ma is currently a JSPS Postdoctoral Fellow in Waseda University, Tokyo. He received the B.S. degree in Mechanical Engineering from Sichuan University, China, in 2009, and Ph.D. degree in Mechanical Engineering from Beijing Institute of Technology (BIT), China, in 2015. His research interests include motion control and human-robot-interaction of humanoid robots, especially involving falling motion control and compliance control.
Title: Damage protection for humanoid robots inspired by human falling
Abstract: In this report, recent studies of humanoid robot falling in Beijing Institute of Technology, China, will be presented. Particularly, we focus on the biomechanical study of human motion. The study of human biomechanics provides useful insight into understanding and properly planning feasible human like motion in humanoids robots. We analyze the motion of human falling based on the motion data acquired from motion capture system. Motion planning of damage protection for humanoid robots are proposed inspired by the human motion. We also design a robot platform for testing and measuring the damage caused by falling down. Experiments of humanoid robot falling showed the effectiveness of our studies.
Libo Meng is currently a researcher in Beijing Institute of Technology, China. His research interests include humanoid robot, biped walking and biomechanical study.