Research On Path Exploration In Unknown Environments By Cooperative Multi-Robot

Exploration in unknown environments, as a fundamental problem in mobile robotics, can be extensively used in planetary exploration,military reconnaissance and disaster rescue. Explorations by coordinated multi-robots have such advantages as parallelism, flexibility, fault tolerance and data redundancy, which not only overcome the uncertainty of the sensors, but also extend the capability of the single robot.Supported by the NSF project,"Indoor Navigation Techniques of Mobile Robot based on Imprecise Map"and the national high-tech research and development plan of China,"Collaboration and Competition Mechanism for Distributed Multi-robots and its Application Techniques", this dissertation aims to improving on the efficiency of the exploration and systemically studies several critical problems existing in cooperative exploration by multi-robots. The main research work is as follows:Firstly, local obstacle avoidance method (reactive navigation) plays an important role in explration by real robot. A variety of velocity space methods such as the curvature velocity method (CVM), the lane curvature method (LCM) and the beam curvature method (BCM) take into account the physical constraints of the environment and the dynamics of the vehicle and formulate the local obstacle avoidance problem as one of constrained optimization in the velocity space. By combining the proposed prediction model of collision with the improved BCM, not only does the method inherit the smoothness of CVM, the safety of LCM and the speediness of BCM, but also it can remedies some shortcomings of the velocity space methods and thus improves the explration efficiency accordingly.Secondly, map building can handle the instinctive limitation that the reactive navigation methods tend to be trapped in local minima and thus improve the exploration efficiency. By taking into account both the advantages and the disadvantages of different environment representations, this dissertation proposes a novel online map building method as well as a novel topological map whose nodes are represented with the range finder's free beams together with the visual scale-invariant features. Compared with the traditional map presentations, neither the accurate global localization of the robot nor the artificial landmarks are required in the map construction. The navigation experiments demonstrate that the topological map is not only easy for construction and maintenance, but also convenient for global pathplaning (deliberate navigation) in large-scale, indoor environment without any landmarks. Therefore, the exploration efficiency can be improved naturally.Thirdly, though the exploration executed by cooperative robots promise several immediate advantages compared to the problems occurring in single robot exploration, the extension to multiple robots brings about several new challenges including local map merging, the selection of cooperation strategy and limited communication. By considering both the similarity of the single node and the spacial relationship between individual nodes, the Hidden Markov Model improves the node's recognition rate greatly and the problem of local map merging is reduced to the fearture matching accordingly. The coorperative exploration experiment demonstrates that the topological map is convenient for the selection of proper coordination strategy so that the time spent on exploration can be reduced.Fourthly, in order to resolve the above mentioned problem of limit communication, the multi-robot system is combined with the wireless sensor networks so that the so called mobile sensor networks can be constructed. The Monte Carlo method that is extensively used in robotics is adopted to resolve the localization of the node so that the Mixture-MCB method is proposed. This method improves on the successful rate of sampling by the mixture sampling mode and therefore resolves the particle degeneration that often happens in traditional Monte Carlo methods. By the deployment of the wireless sensor networks during the course of exploration, the communication burden can be relieved by virtue of the information delivered by the sensors.