Feasibility Verification of Independent Mapping Technology under Semi-Closed Fire Conditions

Abstract

This paper points out that when a fire occurs in some special semi-closed places, the masses cannot quickly and effectively escape from the site safely, propose a method that self-established maps and target recognition through the robot at the site and guides the masses to escape quickly through motion planning. Self-established maps include three parts: self-exploration, simultaneous localization and mapping and motion planning. Self-exploration uses an active exploration method based on information theory to guide the robot to move the maximum information gain between the current point and each point on the two-dimensional map as the target point; simultaneous localization and mapping apply the core idea of particle filtering. The particle group is scattered at the initial position and the weight of each particle is determined by scanning matching method. The point with the largest particle weight is selected as the current position of the robot. Thus constructing of the dimension raster map; The motion planning is divided into global planning and local planning. The global path planning uses the Dijkstra algorithm to find the global optimal path in the two-dimensional grid map. The local path planning uses the DWA (Dynamic Window Approach) algorithm to find the best speed in the local range. Target recognition uses a template matching algorithm to determine the threshold by multiple experiments to achieve accurate target recognition. This paper analyzes and studies the key technologies of robot' s self-construction and target recognition on the turtlebot experimental platform, and completes the technical feasibility verification of the project, so as to reduce the personal injury in the actual scene application. In addition, all processes can be operated and monitored on the remote computer.