Active Contact Engagement for Aerial Navigation in Unknown Environments with Glass

TL;DR

The paper addresses robust aerial navigation in environments with transparent glass by merging incremental visual glass detection with active, contact-based confirmations. A lightweight front-end contact-sensing module and a touch-action planner enable on-the-fly verification of potential glass surfaces, updating a volumetric map and replanning trajectories as needed. Key contributions include (1) an incremental visual glass-detection pipeline with plane fitting and polygonal surface representation, (2) a lightweight flex-sensor contact module with touch-based confirmation, and (3) an autonomous navigation framework that actively engages contacts to safely circumnavigate glass obstacles. Real-world experiments and simulation benchmarks demonstrate improved safety and efficiency over purely non-contact or passive-contact baselines in glass-rich unknown environments.

Abstract

Autonomous aerial robots are increasingly being deployed 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.

Figures (9)

• Figure 1: A real-world experiment photo captured during a touch action, when the aerial vehicle actively contacts with a visually detected glass surface to confirm its presence using a pair of front-mounted contact-sensing modules.
• Figure 2: An illustration of the proposed visual glass surface detection module. (a) Glass segmentation from the color image. (b) Boundary points in the depth image. (c) The 3D point set of the glass surface boundary in the world frame. (d) The polygon of the glass surface on the fitted plane. In (c)(d), the grey voxels are occupancy grids constructed from the depth image.
• Figure 3: The figure illustrates (a) the flex sensor used in our design, (b) the proposed contact-sensing module, (c) the aerial vehicle for flight experiments.
• Figure 4: The figure illustrates the pipeline of the proposed autonomous navigation system, with the pink blocks highlighting components related to active contact engagement.
• Figure 5: An illustration of the proposed autonomous navigation system with active contact engagement in a point-to-point aerial navigation task.