Evaluation of compressive sensing encoding on AR drone

Abstract

Micro-flying robots such as quadcopters or drones are being used extensively in many civilian applications. They are generally integrated with different sensors and are designed to perform tasks both autonomously as well as with manual feedback. These drones typically transmit data periodically either to a base station or to each other if deployed in a swarm. As the number of sensors on-board increases, communication bandwidth becomes a critical aspect for these drones. While there are multiple approaches to improve bandwidth, they typically involve modification of the communication infrastructure. In this paper, we propose a unique method to reduce the data from a typical sensor like an on-board camera using Compressive Sensing (CS) technique. Our method does not require any changes to the communication infrastructure used (WLAN 802.11a) and can be possibly extended to other communication links. We have implemented the CS based encoder on-board the A.R. Drone and conducted various experiments to measure execution time of the processing and quality of the data transmitted to validate our approach.