Learning Adaptive Safety Margins for Visual Navigation

2026-07-20

AI Summary

In complex indoor environments, fixed safety margins often fail to balance collision avoidance safety and path efficiency and are prone to failure due to perceptual uncertainties. This work proposes a context-conditioned safety critic that, for the first time, decomposes adaptive safety margins into three complementary objectives: safety, efficiency, and distance alignment. A two-stage distillation framework enables perception-only safe trajectory selection without explicit planning. The method integrates control barrier functions, ESDF-based geometric priors, diffusion-based trajectory generation, and a teacher–student distillation architecture. Evaluated on the HM3D and MP3D datasets, it achieves state-of-the-art success rate (SR) and success weighted by path length (SPL). Notably, the approach transfers directly to the Unitree G1 humanoid robot without fine-tuning, demonstrating robust navigation in real-world complex scenes.

Abstract

Robots in cluttered indoor spaces often fail not because they cannot generate collision-free paths, but because a fixed safety margin is mis-calibrated: conservative margins cause detours and timeouts, while permissive margins lead to near-boundary shortcuts under perception bias. Diffusion-based planners propose diverse trajectory candidates from egocentric RGB-D, yet reliable selection remains the bottleneck. We propose a context-conditioned safety critic that learns an adaptive clearance preference for ranking diffusion proposals, decomposed into three complementary terms: (i) a safety term with a clearance-budget penalty and a control-barrier-function residual for waypoint- and transition-wise safety, (ii) an efficiency term combining a smoothness penalty with a safety-gated detour-ratio penalty that avoids detours without incentivizing risky shortcuts, and (iii) a distance-constraint matching term that anchors the learned budget to realized ESDF clearances to prevent margin collapse. We train the critic with privileged ESDF geometry in simulation and distill it into a perception-only selector via a two-stage teacher-student procedure. On PointGoal navigation in HM3D and MP3D, including cross-dataset transfer, our method achieves the highest success rate (SR) and success weighted by path length (SPL) among strong diffusion, optimization, and RL baselines. Trained purely in simulation, it transfers to a Unitree G1 humanoid and navigates cluttered indoor scenes without task-specific tuning.

Problem

• visual navigation
• safety margins
• adaptive clearance
• collision avoidance
• perception bias

Innovation

• adaptive safety margins
• diffusion-based planning
• safety critic
• ESDF distillation
• zero-shot sim-to-real transfer