Robot control learns when and how to recover from failure
RAYA: Learning Where and When to Intervene for Robot Recovery
Robotics
Summary
Robots can sometimes predict when they might fail but still fail to stop it in time. The authors found that robots should consider how easy it is to recover while they make decisions, not just react after a problem arises. They created RAYA, a system that learns when and how to adjust tasks to keep the robot safe. Tested in computer simulations and on a small flying drone, RAYA helped the robot complete missions even under tough conditions where other methods failed.
What this means in practice
- •For drone operators: Increase drone mission success under adverse conditions by integrating RAYA for better real-time task adjustment and recovery.
- •For autonomous vehicle teams: Improve failure recovery in autonomous vehicles by using RAYA’s learned intervention and task weighting within vehicle controllers.
Authors
Ishaan Mahajan, Charles Chen, Frederike Dümbgen, Brian Plancher
Abstract
A robot can predict failure and still be unable to prevent it. By the time a safety mechanism reacts, the nominal plan may already have spent the control authority that recovery requires, and fixed task priorities may block whatever response remains. Our key insight is that both aspects are decided inside the controller. Recoverability must inform actions while they are chosen rather than veto them afterward, and task objectives must be adapted as recoverability shrinks. Building on this, we present RAYA, a hybrid learned-analytic framework that places a learned finite-horizon recoverability margin inside an optimal controller with hard constraints and pairs it with a bounded learned scheduler that shifts task weights to facilitate recovery. Across 7,200 simulation episodes per controller spanning quadrotor and autonomous-vehicle benchmarks, RAYA not only improves survival rates, but also transfers the learned components zero-shot to unseen trajectories, disturbances, plant shifts, and friction layouts. We developed an embedded realization of RAYA and deployed it on-board a 35g Crazyflie quadrotor. Across 40 combined hardware flights under wind with either aerodynamic mismatch or an unmodeled 40% motor-command loss, each of three baselines fails in all trials, while RAYA completes 10/10 six-cycle missions. Project Website: https://raya-control.github.io/.