Research On Switching Decision-making And Planning Of Multi-habitat Robots In Complex Environments

Deep research on Air-Ground Amphibious Robot has promoted the development of multirobots with a variety of motion capabilities.Air-Ground Amphibious Robot,as an important branch of amphibious robots,have both air and ground motion capabilities.When the AirGround Amphibious Robot moves autonomously,it is necessary to plan a three-dimensional path without collision for the robot from the starting point to the target point.At the same time,during the movement process,the Air-Ground Amphibious Robot needs to make decisions autonomously based on the current environmental information to select a better motion mode.Therefore,taking the Air-Ground Amphibious Robot in complex environments as the research object,researches on environment modeling,path planning algorithms and switch decisionmaking algorithms have been carried out.First,the basic principle and implementation process of the traditional A * algorithm are explained.When using the grid method for environmental modeling,the height information of obstacles is integrated to build a 2.5-dimensional grid map.While planning,consider the combined cost of motion energy and time consumption in different environments on land and air,and improve the heuristics function of the A * algorithm and remove redundant path points when obtaining a path.When passing the obstacle grid,planning and generating the air path point that is suitable for the motion three-dimensional path of the AirGround Amphibious Robot.Simulation results show that the improved A * algorithm proposed in this paper can shorten the planned path length,reduce the number of turns,and narrow the turning angle.Then,we analyze the ground and air movement modes of the AirGround Amphibious Robot,and establish the energy consumption model under the two modes.When deciding to switch the working mode of the robot,a multi-target optimization decision function is designed in consideration of the energy consumption,time and safety of different modes.By analyzing the environmental state of the Air-Ground Amphibious Robot,a neural network is used instead of the Q-list reinforcement learning algorithm to determine the robot's optimal motion mode in the current environmental state.Simulation and experiments verify that the algorithm can effectively implement the robot's motion mode switch policy decision.Finally,the built Air-Ground Amphibious Robot platform is used for experimental verification.A state machine is designed to realize the robot's autonomous control to switch between the states of the platform.Based on the robot's autonomous control,planning experiments and switching decision experiments are carried out respectively.Experiment results show that the planning algorithm and the decision function applied by the robot can effectively plan the path for the robot according to the environment map and the robot can implement autonomous decision-making based on the decision function.