Research On Group Path Planning Algorithm Guided By Safety Indication Signs

With the rapid development of urbanization and the continuous increase of population,security incidents frequently occur in densely populated areas,and people's lives are threatened.Therefore,it is particularly important to make a reasonable crowd path planning scheme for crowd gathering places.In real life,the safety indication sign is the core part of the crowd evacuation system.Setting a reasonable indication sign can provide the public with effective escape path information.Most of the traditional algorithms only consider to avoid the collision to improve the planning efficiency,but ignore the important guiding role of safety indication signs.Moreover,people are usually regarded as independent individuals to perceive information and make decisions,which does not conform to the real crowd movement behavior.Therefore,in order to make better use of the effective information of the scene for quasi-real path planning,this paper proposes a group path planning method guided by safety indication sign based on the full consideration of the path guiding function of building safety indication sign.This method uses Sarsa algorithm and fuses the group information sharing function to realize the path planning of large groups.In order to further simulate the regional relationship between crowds,the Optimal Reciprocal Collision Avoidance is used to avoid the Collision of local members.Finally,the feasibility and applicability of the proposed method are verified by simulation experiments in different scenarios.The main work of this paper has the following several aspects.1.In view of the lack of utilization of environmental information in the past path planning,this paper proposes a safety indicator sign guidance method.By establishing a guide field of a certain size for the safe indication sign and adding a force with a single direction on the field,the agent will move towards the direction indicated by the safety indication sign once it chooses to enter the guide field.This approach makes the path planning process more consistent with the real situation of crowd movement,and the experimental results prove that the method in this chapter is feasible and can improve the speed of agent path planning to some extent.2.Aiming at the problem of repeated state exploration and information storage redundancy in path planning,this paper proposes a centralized sharing method of environment information.When using reinforcement learning method to plan the path,the agent will store the experience of each time step to the common experience bank,and then periodically selects the experience with the maximum state-action value from the experience bank to learn,so as to complete the use of Shared information.This approach reduces the agent's exploration and learning time to the unknown environment and has a positive impact on the final path planning speed.3.In order to make better use of the guiding function of safety indication sign,this paper proposes a group path planning method based on Sarsa algorithm under the guidance of safety indication sign.In this part,the guidance of safety indication sign is incorporated into the action set as a complete continuous action,and the agent selects the action according to the same reinforcement learning strategy.Taking the environmental perception information and Shared experience information as the input of the state space data of Sarsa algorithm,and using the velocity and angular velocity parameter values as the output to control the motion direction of the agent,the collision-free path planning of a large group is finally completed.Finally,through a lot of experimental analysis and comparison,it is proved that this method can effectively accelerate the speed of crowd evacuation,and can show good expansibility and robustness in different complexity environments.