Research On WiFi Signal Assisted Visual SLAM Method

The classic indoor positioning technology mainly relies on wireless signals such as WiFi signal,electromagnetic,infrared and other to construct a location fingerprint database.Then,utilize the fingerprinting technique to achieve indoor positioning.However,for unknown environments,fingerprinting technique cannot achieve its own location awareness.As mobile robot technology becomes more mature and more widely used,research in the field of robotics related to simultaneous positioning and mapping(SLAM)in unknown and complex environments is becoming increasingly important.Key issue with SLAM technology is closed loop detection.Due to the cumulative error of the robot sensor,the map gradually deviates from the correct trajectory.Therefore,while analyzing some relatively mature SLAM positioning methods,this paper proposed a method for using WiFi signal to assist visual SLAM in terms of closed-loop detection problem.This paper mainly studied the following aspects.Firstly,proposed a method,which in unknown environment WiFi signal assist closed-loop detection.Due to Bag of Words is susceptible to environmental changes,light changes,and obstacle position changes,so this paper identifies the scene based on WiFi signal sequence.Taking advantage of the time and space continuity of WiFi signal,as every time a new key frame is newly inserted,perform a similarity comparison with each existing key frame in the map,then a closed-loop candidate frame set is obtained.According to the Euclidean to judge whether a loop is occurred.Secondly,proposed a method that is constructing a trajectory of a robot using trilateral positioning.The robot monitor WiFi signal strength changes during robot operation.Using the ratio method to realize the trilateral positioning,then get the robot's running track.Thirdly,proposed a method for co-localization of trilateral positioning and visual SLAM.While the robot is running it does not necessarily repeat to a certain position.Therefore,it is impossible to perform closed-loop detection to eliminate system cumulative error.To achieve precise positioning and mapping of visual SLAM,monitorthe WiFi signal strength changes during the operation of the robot.Utilize the triangulation result to construct the scene,correct the SLAM positioning result.Finally,the feasibility and effectiveness of the closed-loop detection method proposed in this paper are verified by experiments.