Multi-Robot Cooperative SLAM
Cooperative SLAM
Click the image for the C-SLAM video
Presentation of Thesis
Click the image to download the ppt
From the early explorers of the Phoenicians to the Mars rovers of NASA, looking for newer turfs and mapping them is something humans have consistently striven for. We have mapped and plotted almost all parts of the visible and traversable Earth. The ones that can be seen from up above have been mapped thanks to the satellites. But some areas like the underwater terrain, deep underground caves, etc. cannot be visited by humans. Exploring other planets and distant celestial bodies by humans also looks like a distant dream because of the safety issues involved, among other things.
Robots used for extra-terrestrial exploration have pre-programmed codes that enable them to make decisions on their own in an unknown, unseen and uncharted environment. One of their primary jobs is to map the surrounding areas to the best of their abilities. With increased number of robots in these operations, not only does the exploration take less time but the maps also become more reliable due to multiple robots’ data. Such robots implement Simultaneous Localization and Mapping techniques (SLAM). Starting from an arbitrary initial point, a mobile robot should be able to autonomously explore an environment with its onboard sensors, gain knowledge about it, interpret the scene, build an appropriate map, and localize itself relative to this map.
SLAM is a major field of study in robotics. The need to automatically and accurately understand a previously unknown environment is finding more and more traction as man ventures more into space exploration. One type of consumer robotic vacuum cleaner is programmed to perform SLAM and create an image of the part of the 2 house it is cleaning. It perceives a part of the house at any given time and plans a local path based on what it sees. Then it moves on to clean other parts of the house.
Autonomous mobile cooperative robotics is what the industry is looking forward to right now. The ability of a robot to move in a confined space, communicate and coordinate with other robots like it and at the same time perform its own duties is highly desirable. Some big industries have implemented such robots in their shop floors and they use it to move light and heavy equipment/materials automatically. In cooperative SLAM, each robot generates its own map of a part of the area which is then compared to the maps developed by other robots. Thus, an important problem in this is to decide when, where and how to merge the different maps of the individual robots. Cooperative SLAM comes from two old ideas: teamwork and curiosity. Entities, human and non-human, working together have been found to complete tasks faster than when they attempt to do them all alone.
One of the inherent challenges in cooperation is constructing a global map from the individual maps generated by each cooperating entity. This thesis presents a solution to the problem of map merging that is simpler to implement than traditional techniques.