Safe & sound: A robust collision avoidance layer for aerial robots based on acoustic sensors

Abstract

In this work, we propose a novel intermediate control layer for obstacle avoidance and drift suppression. It runs in between high level software layers responsible for environment mapping and path planning and low level software ensuring stable and level flight. This allows more complex algorithms with longer computation intervals for higher-level tasks. By modularizing the robot control, we also enable the reuse of our obstacle avoidance layer for different high-layer algorithms. The feasibility of separation in software will be shown, leading the way for a potential physical separation with dedicated hardware later on. A prototypical implementation of the intermediate layer using ultrasonic sensors was built. We demonstrate the feasibility of the chosen sensor concept in presence of propeller noise. The control concept is kept on a level that can be implemented on onboard processors - this does also include the communication with the sensors. Experimental results with this first implemented prototype are presented and discussed. We analyze the ranging capabilities of the ultrasonic sensor and validate its applicability in obstacle avoidance tasks.