aiqbench analysis
RAYA introduces a hybrid learned-analytic framework that integrates recoverability margins directly into optimal controllers, allowing robots to dynamically adapt task priorities before failure occurs rather than reacting afterward. Validated across thousands of simulations and successful on-board deployment on quadrotors under severe disturbances like wind and motor loss, the system demonstrates zero-shot transferability to unseen physical conditions. This approach fundamentally shifts safety from a post-hoc veto mechanism to an intrinsic control objective, significantly enhancing survival rates in autonomous systems.
RAYA: Learning Where and When to Intervene for Robot Recovery
arXiv:2609.21690v1 Announce Type: new 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/.
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: this https URL.
Robotics (cs.RO); Systems and Control (eess.SY)
arXiv:2609.21690 [cs.RO]
(or arXiv:2609.21690v1 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2609.21690 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 295 words
Related coverage
ForkCast: Predictive Recovery for Vision-Language-Action Policies via Counterfactual Futures
arXiv · Oct 14, 2026
REVOLVE: An Automated Closed-Loop Framework for Evolving Robot Manipulation with Minimal Human Intervention
arXiv · Sep 15, 2026
Beyond Failure Recovery: An Engagement-Aware Human-in-the-loop Framework for Robotic Systems
arXiv · Jun 17, 2026
REVERSAL-BENCH: A Reversibility Axis and Reset Oracle for Measuring the Reset-Free RL Cliff
arXiv · Sep 17, 2026