No falls, no resets: Reliable humanoid behavior in the DARPA robotics challenge

By Christopher G. Atkeson, B. W. Babu, Nandan Banerjee, Dmitry Berenson, Christoper P. Bove, Xiongyi Cui, M. DeDonato, R. Du, Siyuan Feng, P. Franklin and 13 more. 2015

114 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

OPAL: Offline Primitive Discovery for Accelerating Offline Reinforcement Learning (2020, background): “For example, a robot learning through trial and error in the real world requires costly human supervision, safety checks, and resets (Atkeson et al., 2015), rendering many standard online RL algorithms inapplicable (Matsushima et al., 2020).”
Deployment-Efficient Reinforcement Learning via Model-Based Offline Optimization (2020, background): “Speciﬁcally, Step 4 requires evaluating the policy for safety, and often requires human monitors (Atkeson et al., 2015).”
Real-Time Perception Meets Reactive Motion Generation (2017, background): “Compared to the aforementioned robot challenges (APC and DRC), our experimental scenarios consider a much smaller variety of manipulation tasks and are tested on a fixed-base platform.”
Reset-free Trial-and-Error Learning for Robot Damage Recovery (2016, methodology): “ost-disaster environments [3], [4]. For instance, C. Atkeson et al. report that the Atlas robot they used in the DARPA Robotics challenge had a “mean time between failures of hours or, at most, days” [1]. The traditional method for damage recovery is to ﬁrst diagnose the failure [5], then update the”