31.12.06 - Bio-inspired Vision-Based Flying Robots. Thèse EPFL, n° 3194 (2005). Dir.: Prof. D. Floreano.

"For his pioneering contribution to the science and technology of autonomous flying micro-robots inspired by the biology of flying insects."

Bio-inspired Vision-Based Flying Robots

Development of small, lightweight, and autonomous flying robots is attracting an increasing interest within the international research community. Such systems have a strong potential in many applications such as aerial surveillance, environmental monitoring, security, ad hoc communications networks, search and rescue, disaster relief, etc. However, no autopilots are yet capable of operating such flying robots in cluttered environments where GPS signals are unavailable and environmental geometry requires continuous and agile obstacle avoidance.

In this thesis, we addressed this problem by constructing a series of small flying robots weighing from 150 to 30 grams and demonstrating autonomous steering capabilities in various indoor environments. In order to achieve this result without using any external beacons nor maps, the sensors equipping those robots as well as their embedded control systems have been largely inspired from the biology of flying insects, which demonstrate dexterous flight control in spite of their tiny brain.

This thesis first dedicates an entire chapter to reviewing the biology of flying insects. It then proceeds with an overview of the robotic platforms that have been developed for testing biologically inspired control systems.
Because flying insects are heavily relying on visual cues and in particular optic flow, one chapter reviews its fundamentals allowing for a better understanding of how images are then processed onboard the flying robots.

Then the proposed control strategies and a characterization of their results are described along with different techniques for their synthesis.