Flying robots are very useful for tasks such as aerial mapping, fast exploration, video footage and monitoring of buildings. However, autonomous missions are usually constrained to high altitude flights above flat surfaces in order to avoid collisions. Traditional methods for collision-free navigation require large resources and limit the payload of robots.
On the other hand, flying insects are able to navigate safely and efficiently using low quality vision as their main sensory input. Ideally, robots could adopt insects strategies, but little is known about the way insects turn visual stimuli into muscular actuation during flight. This precludes a direct implementation of insect flight control into robots and forces roboticists to guess appropriate strategies.
In this thesis, we aim to investigate, develop and implement insect-inspired control strategies in flying robots for goal-oriented navigation in complex environments. We will start by exploring insect sensing and actuation through experiments with bumblebees. Then, we will model the insect’s flight control strategies. This model will be tested in simulation and implemented in a custom designed flying robot. Through experiments in natural environments, we will characterize the efficiency of our model for goal-oriented collision-free navigation.
Spatial Encoding of Translational Optic Flow in Planar Scenes by Elementary Motion Detector Arrays
Scientific Reports. 2018-04-11.
DOI : 10.1038/s41598-018-24162-z.
Miniature artificial compound eyes for optic-flow-based robotic navigation
2014. Workshop on Information Optics WIO2014 , Neuchatel, Switzerland , July 7-11, 2014. p. 1-3.
DOI : 10.1109/WIO.2014.6933290.
A Bee in the Mirror: A Bio-Inspired Model for Vision Based Mid-Air Collision Avoidance
Bionav - The application of animal navigation techniques in autonomous vehicles, Royal Holloway College, Egham, UK, 11-13 April 2013.
A Flying Robot with Adaptive Morphology for Multi-Modal Locomotion
2013. IROS 13 , Tokyo, Japan , November 3-6, 2013. p. 1361-1366.