Learning to Navigate Under Imperfect Perception: Conformalised Segmentation for Safe Reinforcement Learning
This work addresses safety-critical navigation for autonomous systems by offering robust uncertainty handling, though it is incremental in combining existing conformal methods with RL planning.
The paper tackles the problem of safe navigation under imperfect perception by integrating conformal prediction into semantic segmentation to provide finite-sample safety guarantees for hazard detection, resulting in up to 6x increased hazard coverage and up to 50% reduction in hazardous violations during navigation.
Reliable navigation in safety-critical environments requires both accurate hazard perception and principled uncertainty handling to strengthen downstream safety handling. Despite the effectiveness of existing approaches, they assume perfect hazard detection capabilities, while uncertainty-aware perception approaches lack finite-sample guarantees. We present COPPOL, a conformal-driven perception-to-policy learning approach that integrates distribution-free, finite-sample safety guarantees into semantic segmentation, yielding calibrated hazard maps with rigorous bounds for missed detections. These maps induce risk-aware cost fields for downstream RL planning. Across two satellite-derived benchmarks, COPPOL increases hazard coverage (up to 6x) compared to comparative baselines, achieving near-complete detection of unsafe regions while reducing hazardous violations during navigation (up to approx 50%). More importantly, our approach remains robust to distributional shift, preserving both safety and efficiency.