Towards Generalizable Safety in Crowd Navigation via Conformal Uncertainty Handling

2025-08-07

AI Summary

To address the degradation of safety and generalization in mobile robot crowd navigation under out-of-distribution (OOD) scenarios, this paper proposes a novel framework integrating Adaptive Conformal Inference (ACI) with Constrained Reinforcement Learning (CRL). It is the first work to introduce ACI into crowd navigation, dynamically quantifying uncertainty in pedestrian behavior prediction and translating it into real-time soft constraints for policy decision-making—thereby enhancing robustness without compromising safety. Experiments demonstrate that, on in-distribution scenes, the method achieves a success rate of 96.93%, reduces collision frequency by 3.72×, and decreases trajectory intrusion by 2.43×. Moreover, it significantly outperforms baseline methods across diverse OOD scenarios and is validated on a physical robot platform. This work establishes a new paradigm for uncertainty-aware embodied navigation, advancing the integration of statistical reliability estimation with safe, adaptive control in dynamic human environments.

Technology Category

• Intelligent Robots: State Estimation
• Natural Language Processing: Safety and Robustness
• Machine Learning: Calibration & Uncertainty Quantification

Application Category

• Economics, Online Markets and Human Computation: Trust and reliance of crowd workers and data experts on GenAI
• 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

Abstract

Mobile robots navigating in crowds trained using reinforcement learning are known to suffer performance degradation when faced with out-of-distribution scenarios. We propose that by properly accounting for the uncertainties of pedestrians, a robot can learn safe navigation policies that are robust to distribution shifts. Our method augments agent observations with prediction uncertainty estimates generated by adaptive conformal inference, and it uses these estimates to guide the agent's behavior through constrained reinforcement learning. The system helps regulate the agent's actions and enables it to adapt to distribution shifts. In the in-distribution setting, our approach achieves a 96.93% success rate, which is over 8.80% higher than the previous state-of-the-art baselines with over 3.72 times fewer collisions and 2.43 times fewer intrusions into ground-truth human future trajectories. In three out-of-distribution scenarios, our method shows much stronger robustness when facing distribution shifts in velocity variations, policy changes, and transitions from individual to group dynamics. We deploy our method on a real robot, and experiments show that the robot makes safe and robust decisions when interacting with both sparse and dense crowds. Our code and videos are available on https://gen-safe-nav.github.io/.

Problem

• Enhancing robot crowd navigation safety under distribution shifts
• Using conformal uncertainty to guide robust reinforcement learning
• Improving performance in out-of-distribution pedestrian scenarios

Innovation

• Uses conformal inference for uncertainty estimates
• Applies constrained reinforcement learning for safety
• Adapts to distribution shifts robustly