Quadcopter Obstacle Avoidance using Biomimetic Algorithms

Unmanned Micro Aerial Vehicles (MAVs) have the potential to operate in diverse environments but are limited by the lack of robust algorithms for autonomous flight. This is largely due to the sensing and processing requirements that exceed the weight and power limitations of this hardware. Recent research has highlighted the potential to overcome these constraints by looking to the natural world, in particular to the possibilities of using optical flow. This work presents a novel biomimetic algorithm that uses optical flow data generated from the on-board camera of a quadcopter MAV to avoid obstacles in flight. Simulation results are presented showing the algorithm performance in a range of flying scenarios. This work also highlights the huge potential of using low resolution sensors and lightweight algorithms in the field of autonomous vehicle control.