Research On Key Technologies Of Multiple Autonomous Mobile Robot Cooperation

Multi-robot cooperation technology research has become one of the most potentialdirections of robot technology. With the development of scientific research, in order to replacehuman and obtain the first-hand information of various dangerous environment, multi-robotcooperation has become a necessity technology for scientific exploration. This dissertationmainly studies the system structure, cooperation positioning and task allocation of multi-robotcooperation technology, the main work as follows:According to the development of the autonomous mobile robot technology, thisdissertation makes a detailed introduction on practical application of the technologies of robotcooperation. From the development of general thinking and trends of robot technology, itelaborates the multi-robot cooperation as the inevitable outcome and goals of the developmentof robot technology. In this dissertation an summary was made about the key technology,difficulties, the existing problems of the multi-robot cooperation. In addition, a briefintroduction about the research foundation of this paper was provided.The dissertation studies the multi-robot cooperative positioning. The key technology ofcooperative positioning is information communication between robots and to get globalinformation from local information fusion. For the problem of information fusion, thisdissertation adopts a kind of distributed multi-sensor information fusion method to deal withthe information fusion between odometer and visual sensor, meanwhile matching the initialfusion result through the Mahalanobis distance data correlation model, which improves thereliability and fault tolerance of the system. A kind of extended Kalman filter positioningmethod based on feature detection was put forward. In this way,through the relativeobservation information of the robot,we are able to update the robot’s equation of state andobservation,which can improve the accuracy of position. Finally, we give positioning error ofthe simulation. It is proved reliable and effective.The dissertation does some research on the distribution multi-robot task. Firstly, wesummarize and classify the current main methods of task allocation and describe the taskallocation problem in mathematics method. Multi-robot task allocation problem belongs to thecategory of the robot’s upper control system, which needs dealing with hierarchicalcoordination. Through the description about the state-space of multi-robots and multitasking,we can makes a description to multi-robots and multitasking in the form of a finite set respectively. So the robot task allocation can be seen as a combinatorial optimization problemwhich has constraint conditions. The dissertation puts forward the definition of ability vectorto define robot and task in the form of ability and make sure the conditions in which robotscan complete their tasks. Market mechanism is an effective method to have a reasonableallocation of existing resources,which can come up with a dynamic multi-robot taskallocation method based on market mechanism combined with the functional structure ofmulti-robot task allocation system. We can get the relationship between the number of robotsand the number of tasks with this allocation through the analysis of simulation results. Finally,the feasibility of the method was demonstrated through the dynamic role assignment in robotsoccer game.