Pose Correction of Autonomous Vehicles with Edge Computing

Although people have made great progress in the field of artificial intelligence such as driverless cars, it is quite challenging for artificial intelligence system to detect the errors caused. Once these undetected errors accumulate to some extent, the system would collapse, thus causing severe consequences. Though loop closing is effective in identifying and correcting cumulative error, it is suitable for the situations where loops exist. The development of edge computing and intelligent network technology opens up new possibilities to address these problems. In this paper, a new framework is proposed on the basis of edge computing and intelligent network to correct the cumulative error of the autonomous vehicle. Under this framework, the autonomous vehicle is capable to calculate and correct the cumulative error based on the information provided by the edge platform. A general and simple realized error correction algorithm is proposed. Experimental results show that this framework reduce the positioning cumulative error of the driverless car effectively. The observation accuracy and period, and the system delay are proved to influence the final accuracy. Finally, some suggestions are made.