Cerebellar-Inspired Residual Control for Fault Recovery: From Inference-Time Adaptation to Structural Consolidation

2026-02-06

AI Summary

This work proposes a cerebellum-inspired residual control framework to address the challenge of online adaptation in robotic policies following deployment under actuator failures, dynamic changes, or environmental perturbations. The approach enhances a frozen reinforcement learning policy during inference by applying online action corrections without altering the original policy parameters, thereby enabling fault recovery. Key innovations include high-dimensional pattern separation, parallel microzone residual pathways, multi-timescale local error-driven plasticity, and a conservative meta-adaptation mechanism that balances recovery speed with behavioral stability. Evaluated on MuJoCo benchmarks, the method improves performance by 66% and 53% for HalfCheetah-v5 and Humanoid-v5, respectively, under moderate faults, while exhibiting graceful degradation under severe disturbances. Further robustness is achieved through residual architecture consolidation.

Technology Category

• Intelligent Robots: Behavior Learning & Control
• Machine Learning: Adversarial Learning & Robustness
• Planning, Routing, and Scheduling: Replanning and Plan Repair

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems

Abstract

Robotic policies deployed in real-world environments often encounter post-training faults, where retraining, exploration, or system identification are impractical. We introduce an inference-time, cerebellar-inspired residual control framework that augments a frozen reinforcement learning policy with online corrective actions, enabling fault recovery without modifying base policy parameters. The framework instantiates core cerebellar principles, including high-dimensional pattern separation via fixed feature expansion, parallel microzone-style residual pathways, and local error-driven plasticity with excitatory and inhibitory eligibility traces operating at distinct time scales. These mechanisms enable fast, localized correction under post-training disturbances while avoiding destabilizing global policy updates. A conservative, performance-driven meta-adaptation regulates residual authority and plasticity, preserving nominal behavior and suppressing unnecessary intervention. Experiments on MuJoCo benchmarks under actuator, dynamic, and environmental perturbations show improvements of up to $+66\%$ on \texttt{HalfCheetah-v5} and $+53\%$ on \texttt{Humanoid-v5} under moderate faults, with graceful degradation under severe shifts and complementary robustness from consolidating persistent residual corrections into policy parameters.

Problem

• fault recovery
• post-training faults
• robotic policies
• inference-time adaptation
• real-world deployment

Innovation

• cerebellar-inspired control
• inference-time adaptation
• residual policy
• error-driven plasticity
• fault recovery