Egocentric Tactile and Proximity Sensors as Observation Priors for Humanoid Collision Avoidance

2026-04-28

AI Summary

This work investigates the design of tactile and proximity sensors—including their modality, coverage, and sensing range—to enable efficient whole-body collision avoidance for humanoid robots. Leveraging a reinforcement learning framework, the study systematically evaluates the impact of various sensor configurations on learning efficiency and obstacle avoidance performance using a dodgeball task as a benchmark on the H1-2 humanoid platform. The findings reveal that, given sufficient sensing distance, raw proximity signals can effectively replace explicit object localization, and that sparse, non-directional proximity sensing yields higher sample efficiency than dense, directional alternatives. Through sensor ablation studies and multimodal perception fusion, the work quantifies the influence of key sensor attributes on policy learning and demonstrates an effective whole-body collision avoidance control strategy.

Abstract

Collision-free motion is often aided by tactile and proximity sensors distributed on the body of the robot due to their resistance to occlusion as opposed to external cameras. However, how to shape the sensor's properties, such as sensing coverage; type; and range, to enable avoidant behavior remains unclear. In this work, we present a reinforcement learning framework for whole-body collision avoidance on a humanoid H1-2 robot and use it to characterize how sensor properties shape learned avoidance behavior. Using dodgeball as a benchmark task, we ablate the properties of sensors distributed across the upper body of the robot and find that raw proximity measurements can substitute for explicit object localization provided the sensing range is sufficient and that sparse non-directional proximity signals outpace dense directional alternatives in sample efficiency.

Problem

• egocentric sensors
• collision avoidance
• sensor properties
• humanoid robot
• proximity sensing

Innovation

• egocentric sensing
• reinforcement learning
• collision avoidance
• proximity sensors
• humanoid robot