A closed loop detection method for lidar simultaneous localization and mapping with light calibration information

Abstract

In this work, we study the loop detection problem for mobile robots in the process of simultaneous localization and mapping. In a complex environment, the accuracy and recall rate of loop detection will decrease significantly because of the lack of prominent environmental state information. The proposed method combines light fidelity (Li-Fi), used for sensing the position of mobile robots, with laser ranging technology to recognize the location that the robot has been aware of during the period of locating and mapping. A shortest link processing mechanism for Li-Fi calibration is proposed to select data, thus smoothing the system. Using an extended Kalman filter, we construct virtual consistent state components of Li-Fi information. Scans are enriched by providing more environmental information in this method. A distanceregion model is designed to reduce the amount of calculation appropriately in the process of scan matching. The transformation between frames is reconstructed using a weight distribution from the data of laser and photoelectric sensors to reduce the error of data association. To reduce the impact of Li-Fi link occlusion, we set up a pause mechanism during the period of fusing information in the process of scan-to-map matching. Results of experiments in corridors and offices show that the proposed method can detect key frames, effectively suppress pose drift, improve the accuracy, and guarantee a satisfactory recall rate of loop closure detection.