Pedestrian Flow System Modeling And Robot Control Algorithm Research

Pedestrian flow system exists widely in real life,so it is necessary to flexibly control the pedestrian flow at the entrance and exit of the passage.In order to save the expensive setting and renovation cost of infrastructure such as gate machines,this paper proposes a robot-assisted pedestrian flow regulation scheme and analyzes the passive human-robot interaction(HRI)process,in which the robot is regarded as a dynamic obstacle interacting with the pedestrian.To realize this scheme,this paper respectively studies the robot path planning algorithm,human-computer interaction process and pedestrian flow modeling and control algorithm,and uses ROS based wheeled robot for experimental verification.Firstly,the path planning problem of robots in complex environments is studied,and a smooth A~* algorithm based on map compression is proposed,which improves the problem that traditional A~* algorithm cannot consider the robot radii and have too many transitions.An improved DWA algorithm based on Qlearning is proposed,the evaluation functions of the traditional algorithm are improved and added.The Qlearning algorithm is applied to the DWA to achieve parameter self-adaptation,which enhances the performance of path finding and obstacle avoidance.Combining the advantages of the two improved algorithms,a hybrid algorithm is designed to realize path planning in complex environments.Secondly,the passive human-computer interaction motion state in pedestrian passage is studied,and a pedestrian passage HRI motion simulation algorithm based on improved artificial potential field method for is proposed,where the robot is regarded as a dynamic obstacle placed in the pedestrian flow.In order to obtain the real pedestrian trajectory,a new distribution form of force field and dynamic state field is proposed to solve the problems of the traditional artificial potential field method,such as unsatisfactory velocity,trajectory and poor dynamic obstacle avoidance performance.The potential field model in the pedestrian passage is established to conduct HRI motion simulation under different robot motion modes,and the relationship between robot motion speed and pedestrian speed is fitted out.Then,the motion state of pedestrian flow in T-shaped channel is studied,and a macroscopic modeling and control algorithm of pedestrian flow is proposed.For the T-shaped pedestrain channal with several controlled entrances,the macroscopic model of pedestrian motion state is established with and without considering the confluence node.Two kinds of controllers are designed to make the pedestrian system asymptotically stable and the traffic efficiency optimized.The value range of control parameters and the expected flow rate of each controlled entrances are obtained.The effectiveness of the control is proved by theory and simulation.Finally,the path planning experiment and HRI experiment of wheeled robot are carried out.The software and hardware platform are builted.The simulation software and real corridor are used for robot path planning experiments to verify the effectiveness of the algorithm.Passive human interaction experiment is carried out in pedestrian channels to obtain the relationship between robot motion speed and pedestrian speed,which verifies the feasibility of HRI simulation results and robot-assisted pedestrian flow control...