| Robot has played an increasingly important role in target search and scene modeling in unknown environments such as reconnaissance missions,deep-sea exploration,disaster rescue,search and rescue.In some unknown danger scenarios,the use of robot instead of human exploration can not only avoid possible casualties,but also achieve fast and accurate environmental modeling.In the face of increasingly complex environments and tasks,independent single-robot operations are increasingly inadequate,and it is no longer possible to meet people’s objective needs,and it is necessary to use multi-robot collaborative work.In thesis,the two tasks of multi-robot collaborative exploration of unknown environment and on-duty monitoring of known environment are studied,in which obtaining unknown environment information is the precursor task of various subsequent derivative tasks,which can provide maps for subsequent tasks.Static on-duty can obtain information about critical areas in real time.The main work of this article is as follows:1.Thesis analysis and summarizes the traditional multi-robot collaborative exploration framework based on 2D map,and compares the advantages and disadvantages of multiple collaborative allocation algorithms in multi-robot collaborative exploration tasks through simulation experimental analysis.Aiming at the problem of navigation stuck in actual complex scenarios,thesis proposes to determine whether there is a passable path between the robot and the target point,filter the unreachable target point,and verify through physical experiments to solve the navigation stuck problem.2.Thesis focuses on the research of multi-robot collaborative exploration of unknown environments based on 3D maps,and develops multi-robot collaborative exploration algorithms M-TARE with low communication requirements and low computing requirements based on the current best single-robot exploration algorithm TARE＿Planner(Technologies for Autonomous Robot Exploration Planner)as the basic component.1)The initial calibration and SLAM(Simultinous Localization And Mapping)algorithm are used to solve the problem of multi-coordinate system;2)Maintain a global mesh under a unified coordinate system to share information about the explored area;3)Distributed decision-making is adopted,and the collaborative allocation of the area to be explored is modeled as MDVRP(Multi Depot Vehicle Routing Problem);4)Two kinds of "simple map accessible by multiple robots" are innovatively proposed for calculating travel distance and far point navigation.Through simulation experiments,the M-TAREMerger scheme based on global merge map is verified,the average single planning time is maintained at 0.5s;the exploration time of a single robot is the same as that of TARE＿Planner,and the exploration efficiency of two robots and three robots is 1.68 x and2.25 x of that of a single robot.The communication requirement of three robots based on global merge map is only 3KB/s.Real experiments were carried out on a dual vehicle equipped with the edge onboard RK3588 S.It is confirmed that the multi-robot collaborative exploration algorithm has low communication requirements and low computing requirements.3.Thesis develops a multi-robot static on-duty scheme based on 2D map,the conventional solution space of static duty is the entire map,and thesis reduces the solution space of the duty by introducing the "main road" of the environment,from the order of magnitude of the entire map to the order of magnitude of multiple paths.Simulation results show that the proposed algorithm effectively reduces the solution complexity. |
