Research On Complete Coverage Of Autonomous Mobile Robot And Multi-task Allocation

With the fast development of new principle in intelligent control, computer network and artificial intelligence, mobile robot and multi-robot system has become a focal problem in the field of robotics and automation. Represented by cleaning robots, vacuum robots and lawnmowers, complete coverage path panning, which is a pivotal part of the intelligent behavior, has a broad prospect of application and gradually becomes an important new focus. Multi-task allocation is a basic problem of multi-robot system, which reflects the high-level forms of organization and operating mechanism. It is the basic goal of multi-robot system implementation. Complete coverage path palnning problem of nonholonomic wheeled mobile robot (WMR) under unknown static area and multi-robot task allocation problem are deeply studied as the key contents respectively. Meanwhile, this dissertation presents appropriate solutions.As for complete coverage problem, this paper proposes a new strategy. For the task of covering whole region with obstacles, this strategy takes all three subproblems such as region decomposition, complete coverage of subregion and subregion linking into account together to achieve the overall optimization. Boustrophedon decomposition is used for region decomposition; Roundabout sweeping with forward steering and stream function based on path following to avoid unknown static obstacles are adopted to complete coverage of subregion; subregion linking is realized by point stabilization based on fuzzy logic control. Results simulated on the platform of MRS launched by Microsoft Corporation show that it is very suitable for simple robot. Besides this, it has less information storage, efficient, and even adapt for complex environment.As for multi-task allocation problem, this paper researchs the TSP best traversal based on ant colony algorithm and two models of multi-task allocation. We propose a new Adaptive Ant Colony Algorithm (AACA) improved traditional ant colony algorithm and test on three fundamental ISP problems (Eil51, Eil76and CHN130). The simulation results show that AACA is proved in expanding the search of global optimal solution and convergence speed. Finally, AACA is expanded to the application of two kinds of models of multi-robot task allocation (ST-MR-IA and MT-MR-TE) to solve the problem of high-level task allocation and it is verified to be feasible and effective.