Online Mapping and Motion Planning under Uncertainty for Safe Navigation in Unknown Environments

Safe autonomous navigation is an essential and challenging problem for robots\noperating in highly unstructured or completely unknown environments. Under\nthese conditions, not only robotic systems must deal with limited localisation\ninformation, but also their manoeuvrability is constrained by their dynamics\nand often suffer from uncertainty. In order to cope with these constraints,\nthis manuscript proposes an uncertainty-based framework for mapping and\nplanning feasible motions online with probabilistic safety-guarantees. The\nproposed approach deals with the motion, probabilistic safety, and online\ncomputation constraints by: (i) incrementally mapping the surroundings to build\nan uncertainty-aware representation of the environment, and (ii) iteratively\n(re)planning trajectories to goal that are kinodynamically feasible and\nprobabilistically safe through a multi-layered sampling-based planner in the\nbelief space. In-depth empirical analyses illustrate some important properties\nof this approach, namely, (a) the multi-layered planning strategy enables rapid\nexploration of the high-dimensional belief space while preserving asymptotic\noptimality and completeness guarantees, and (b) the proposed routine for\nprobabilistic collision checking results in tighter probability bounds in\ncomparison to other uncertainty-aware planners in the literature. Furthermore,\nreal-world in-water experimental evaluation on a non-holonomic torpedo-shaped\nautonomous underwater vehicle and simulated trials in the Stairwell scenario of\nthe DARPA Subterranean Challenge 2019 on a quadrotor unmanned aerial vehicle\ndemonstrate the efficacy of the method as well as its suitability for systems\nwith limited on-board computational power.\n