Research On The Navigation System Of Fire Inspection Robot

With the widespread application of robots,unmanned areas in manufacturing and daily life are rapidly increasing.An inevitable issue that manufacturers always concern for some complex unmanned manufacturing areas is the dangerous and circulating production may result in potential safety hazard and it may even cause catastrophic failures.In order to meet the requirements of fire safety in production,a type of fire inspection robot that can be on duty around the clock and can conduct fire safety inspections of the unmanned area is needed.Aiming at the characteristics of the working environment of the fire inspection robot,this paper proposes a novel navigation system for the fire inspection robot based on the fusion of binocular vision system and lidar information.According to the characteristics of strong liquidity and frequent environmental changes in the working environment of the fire inspection robot,a binocular vision system is constructed to collect environmental images to obtain more detailed environmental information.For the images collected by the binocular vision system,a binocular vision odometer based on ORB feature point extraction and matching is designed.The EPn P algorithm is used to solve the relative pose of the robot in the environment based on the feature points in the environment.According to the environmental feature points and robot paths obtained by the visual odometer,the ORB-SLAM algorithm is utilized to construct a sparse skeleton map of the environment.However,the sparse skeleton map constructed in traditional ORB-SLAM describes the environment very briefly and is difficult to be applied to the guidance of subsequent path planning.For this issue,an extended ORB-SLAM algorithm that combines binocular vision and lidar is proposed,in which the quaternion method is introduced to fuse the dense point cloud data obtained by lidar with the environmental sparse feature point cloud obtained by the binocular vision odometer,and a dense map suitable for navigation applications is filled subsequently based on the sparse skeleton map.In order to verify the designed navigation system for the fire inspection robot,a fire inspection robot experiment platform based on the wheeled robot chassis is built.The platform is equipped with a binocular vision system and lidar for the needs of the fire inspection robot.The navigation system is validated in a preset environment simulating the actual working environment of the fire inspection robot.The experimental results prove that the designed navigation system can meet the working requirements of the fire inspection robot.