Target Localization and SLAM for UAVS
Gianpaolo Conte and Patrick Doherty
This project focuses on the development of techniques that enable one or several small Unmanned Aerial Vehicle (UAV) platforms to localize one or several targets (ground-based objects) in the environment. Several potential scenarios will be analyzed.
As first scenario, the position of the UAVs is assumed to be known (using GPS in outdoor operation) while the target's position will be unknown. Different target localization techniques will be analyzed and compared. The possibility of using several cooperative UAV platforms as a way to improve the estimation of target geo-location will be analyzed. In addition, the possibility of using pre-existing geo-reference imagery (satellite or aerial images) for accurate target collocation through image matching techniques will be investigated.
As second scenario, the position of the UAVs and targets are unknown. In this case the UAVs have to localize themselves and the targets in the environment. This problem falls into the general Simultaneous Localization And Mapping (SLAM) problem. SLAM is a challenging research problem, especially for UAVs. This technique potentially can solve the navigation problem of a UAV flying in environments with poor or no GPS coverage (for example in indoor environments).
As third scenario, a micro UAV has to localize a target in an unknown indoor environment. This scenario is essentially experimental and a human operator will play a central role in solving the task. A forward looking on-board camera will be placed on a micro rotorcraft. A human operator has the task to fly the rotorcraft using goggles and search for targets using only the visual feedback from the on-board camera. This challenging scenario will emphasize the problems related to the interaction between human pilot and UAV operating in an unknown cluttered environment. Although for a human pilot, the task of operating a UAV through goggles is relatively easy in a large unconstrained outdoor environment, it becomes a great challenge when the environment is constrained and cluttered with obstacles. The challenge is due to the fact that the pilot has only limited perception capabilities (camera field of view). Different ways to enhance the pilot's perception of the UAV's environment will be investigated.
The platforms which will be used in this project are the micro rotorcraft LinkMAV (Fig.1) designed and developed in the AIICS division at the Department of Computer and Information Sciences, Linköping University, and the RMAX autonomous helicopter (Fig.2), an industrial grade UAV produced by the Yamaha Motor Company. The RMAX avionics has also been developed in the AIICS division using off-the-shelf hardware components.