Active Contact Engagement for Aerial Navigation in Unknown Environments with Glass
Xinyi Chen, Yichen Zhang, Hetai ZOU, Junzhe WANG, Shaojie Shen
AI summary
Problem
Transparent glass obstacles cause non-contact sensors like cameras and LiDAR to fail, while existing contact-resilient drone designs are often heavy, complex, or rely on inefficient passive collisions.
Approach
The system incrementally detects potential glass surfaces using onboard RGB-D cameras, then actively guides the drone to gently touch them with lightweight flex sensors to confirm their presence, updating a volumetric map and replanning trajectories in real-time.
Key results
- Incremental visual glass detection method that extracts and maintains 3D surface information from RGB-D frames
- Lightweight 3.3 g flex-sensor module enabling reliable touch confirmation without hard collisions
- Autonomous navigation pipeline that actively plans gentle touch trajectories to validate glass and update maps on the fly
- Successful real-world validation in glass-rich environments demonstrating safe, robust point-to-point navigation
Why it matters
Enables safer and more reliable autonomous drone operations in real-world settings like buildings and industrial sites where transparent glass is common.
Abstract
Autonomous aerial robots are increasingly being de- ployed in real-world scenarios, where transparent glass obstacles present significant challenges to reliable navigation. Researchers have investigated the use of non-contact sensors and passive contact-resilient aerial vehicle designs to detect glass surfaces, which are often limited in terms of robustness and efficiency. In this work, we propose a novel approach for robust autonomous aerial navigation in unknown environments with transparent glass obstacles, combining the strengths of both sensor-based and contact-based glass detection. The proposed system begins with the incremental detection and information maintenance about potential glass surfaces using visual sensor measurements. The vehicle then actively engages in touch actions with the visually detected potential glass surfaces using a pair of lightweight contact-sensing modules to confirm or invalidate their presence. Following this, the volumetric map is efficiently updated with the glass surface information and safe trajectories are replanned on the fly to circumvent the glass obstacles. We validate the proposed system through real-world experiments in various scenarios, demonstrating its effectiveness in enabling efficient and robust autonomous aerial navigation in complex real-world environments with glass obstacles.