Radar Classification Of Obstacles in Nature

Forests cover more than 40% of Europe’s surface and are essential for biodiversity, provide fresh water, absorb carbon and preventerosion. Yet they face detrimental effects of climate change, such as wildfires or outbreaks of the bark beetle. The field of roboticsoffers a pallet of tools to help manage and monitor forests, yet mainly by flying robots.

Ground robots that could carry heavierequipment and last longer struggle in vegetation since their autonomy systems have been developed for obstacle-free scenarios(e.g., driving on roads).

The research proposed here, “Radar Classification Of Obstacles in Nature (RaCOON)”, aims to enable thedeployment of ground robots in forests by giving them the ability to decide which vegetation can be safely driven through.

Theapplicant will deploy a new sensor modality, i.e. radar, and develop a novel sensor fusion system that will classify vegetation into theobstacle and non-obstacle categories. This additional information will allow ground robots to autonomously plan trajectories andnavigate in vegetation.

The problem will be approached first by exploring the possibilities of radars in a proof-of-concept experiment.Then, a forest robotic dataset will be recorded in various types of vegetation.

The experience from the proof-of-concept experimentand the recorded data will motivate the design of the final sensor fusion system.

The outcomes of RaCOON will be 1) dissemination ofthe new system and dataset to the research community and professional networks, 2) training of the applicant in the deployment ofradars for mobile robots and 3) extending the applicant’s professional network and independent research capabilities, advancing him towards starting his own robust field robotics research group.