Ravi Kulan Rathnam

• Exploration can be termed as the process of autonomously navigating in an environment in order to build a map. This is one of the core components for autonomous robots. Mobile robots are increasingly being used in areas where human interaction is not present or extremely limited. These include realms such as surfaces of planets and moons, underwater sites, nuclear power plants etc. In such situations, robots need to navigate the environment without human intervention. They need to decide the next goal points to reach in order to increase their knowledge of the environment, using just the information collected so far. With great advances being made in the field of single robot mapping and exploration, the next step is to use multiple robots. The obvious advantage is the speed up of exploration time by distributing the exploration task among multiple robots. Furthermore, the use of multiple robots ensures that there is no single point of failure, increasing redundancy, allowing for increased robustness. Also, multiple robots may help in localisation by detecting each other, or the same feature from multiple locations. Most of the work done till now has focused on 2D exploration. This thesis proposes an exploration method which performs true 3D exploration, with robots navigating in a 3D environment to produce a 3D map. Furthermore, the members of the robot team stay within communication range of each other, thereby always being able to communicate, and hence being able to cooperate in the exploration process. Also, multi-robot exploration has a very large search space. This means that it is not possible to arrive at an optimal solution in a reasonable amount of time. This requires sampling the search space to reach a solution. The random nature of the algorithm often leads to suboptimal solutions. In this thesis, the random sampling is augmented by a heuristic which improves the exploration time by preventing unnecessary steps during the exploration process. The algorithm is tested in a high-fidelity simulator which was developed during the course of the project which takes into account the dynamics of the robot, thereby allowing a seamless transition to real life systems.