Research On SLAM Navigation System Of The Inspection Robot For Workshop

After "Made in China 2025" was put forward,China implemented the strategy of advancing the manufacturing power and promoted the transformation and upgrading of the manufacturing industry.The development direction of intelligent manufacturing to deeply integrate industrialization and informatization is very necessary for the digitalization and structuring of the production process.For traditional manual inspections,the quality of inspections is difficult to guarantee,and there are problems such as missed inspections,on behalf of inspections,and non-compliant inspections.With the continuous breakthrough of robot intelligent technology,robot technology is used in workshops instead of manual inspections to realize intelligent inspections.Inspection helps to solve traditional inspection problems.Traditional navigation technologies,such as magnetic navigation and two-dimensional code navigation,all need to modify the venue,and there are the disadvantages of fixed motion paths and inability to avoid obstacles autonomously during navigation.Inspections in the workshop environment require more widely applicable technologies.SLAM navigation can achieve accurate autonomous positioning and autonomous obstacle avoidance with variable paths without changing the environment,so that the workshop inspection robot can cope with the unstructured workshop environment where dynamic and static environments coexist.The key to the realization of SLAM navigation lies in the positioning and obstacle avoidance autonomous technology.Inaccurate and reliable positioning will cause the robot to obtain the wrong state and lead to subsequent planning and avoidance errors.Ineffective autonomous obstacle avoidance will cause the robot to be trapped and unable to smoothly bypass the obstacles to perform tasks..In order to realize the autonomous and intelligent inspection of workshop inspection robots,this paper conducts related research on the SLAM navigation system of workshop inspection robots and its key technologies.The main research contents are as follows:(1)Model the robot SLAM navigation system.The overall framework of the workshop inspection robot navigation system is proposed,and the underlying model and navigation map model of the multiple sensors used are analyzed to provide a theoretical basis for localization of combined filter using multi-sensor fusion and autonomous obstacle avoidance under 3D perception.(2)Considering the local defects of a single sensor,this paper proposes a combined filtering localization method based on multi-sensor fusion.State estimation of multiple sensor data,combined use of Kalman filter and particle filter step-by-step fusion,and combined with the hybrid map model to improve the robot's local localization and global localization performance,the average absolute trajectory error of the local localization during the navigation process is about 0.061 m,the effective average localization error of the global localization is 0.098 m,the average attitude error is 1.413°,and the average matching degree is about 76.7%.Compared with the single sensor method,the effect is more accurate and stable.(3)Since lidar only scans at a fixed height,it will scan shelves or pedestrians incompletely.This paper proposes a 3D perception method that combines depth camera and laser for perception processing,and tests and compares DWA and TEB autonomous obstacle avoidance algorithms to combine the overall autonomous obstacle avoidance method.Experiments have verified that the three-dimensional perception method in this paper can perceive and supplement the obstacles on the shelf in terms of perception.Although the TEB algorithm selected in the planning aspect will fall into the local optimum like DWA,it can be separated from the local optimum through a period of optimization.Planning The performance is better,in terms of the overall aspect,it can effectively realize a more complete perception of obstacles,assisting local planning can directly bypass when encountering obstacles without falling into the local optimum,and realize autonomous obstacle avoidance.(4)Build the software and hardware system of the robot platform,and in consideration of closer to the actual application scenarios,select a larger site,design more challenging navigation conditions,and verify the versatility and stability of the above localization and obstacle avoidance methods.Finally,use the patrol instrument reading to test the overall accuracy and detection effect of the robot.Experiments have verified that the localization and stability of the system can achieve the desired effect,and the average position error is about3.81 cm and the average attitude error is about 3.96° during the instrument detection.The camera effect can be directly on the instrument and accurate readings are performed,which proves the navigation system The feasibility and effectiveness of inspections.