When Your Robot Breaks: Active Learning During Plant Failure
This addresses the critical issue of robot resilience and safety during failures, which is important for applications like autonomous vehicles and drones, though it appears incremental as it builds on existing techniques like model predictive control and active learning.
The paper tackles the problem of enabling robots to quickly and safely learn new dynamics after catastrophic failures, such as physical damage, by proposing a probabilistically-safe, online learning framework that combines model predictive control and active learning. The result is a method that can regain control of a severely damaged aircraft in seconds and find safe, information-rich trajectories in only 0.1 seconds, outperforming state-of-the-art approaches.
Detecting and adapting to catastrophic failures in robotic systems requires a robot to learn its new dynamics quickly and safely to best accomplish its goals. To address this challenging problem, we propose probabilistically-safe, online learning techniques to infer the altered dynamics of a robot at the moment a failure (e.g., physical damage) occurs. We combine model predictive control and active learning within a chance-constrained optimization framework to safely and efficiently learn the new plant model of the robot. We leverage a neural network for function approximation in learning the latent dynamics of the robot under failure conditions. Our framework generalizes to various damage conditions while being computationally light-weight to advance real-time deployment. We empirically validate within a virtual environment that we can regain control of a severely damaged aircraft in seconds and require only 0.1 seconds to find safe, information-rich trajectories, outperforming state-of-the-art approaches.