Distributed Localization and Mapping for Aerial Vehicles
Simultaneous localization and mapping (SLAM) has been an active research area in ground robotics for a number of years with impressive results. Localization is the problem of finding out a robotís coordinates relative to its environment assuming one has a map of the environment. Mapping is the problem of acquiring spatial models of the robotís environment which implies knowing the robotís coordinates. SLAM is the problem of doing both simultaneously and is considered one of the key functionalities required for truly autonomous robots. Much of the previous work has focused on environments which are static, structured and of limited size using single ground robotic systems as the platform of choice.
Our integrated MOVIII project will focus on the SLAM, localization, mapping and navigation problems in the context of aerial robotic platforms. An additional dimension will attempt to leverage the use of several aerial vehicles for map building using a distributed, cooperative approach. The problems proposed provide additional challenges in this context since the environments are dynamic, relatively unstructured and often of very large size. In addition, aerial robots have a much faster dynamics as compared with ground robots.
Milestone 1: Automated return mission (18 months). The core techniques of automated detection and recognition of natural landmarks from vision, and fusing this with low-accuracy (affordable) inertial data will be developed to enable automated landing at home base. Yamaha RMAX helicopter platforms will be used.
Milestone 2: Cooperative SLAM mission: (36 months). A couple of UAVís will leave a base with the mission to map the local neighborhood and find a hidden object.