Implementation of a Server-based Multi-Client Autonomous Navigation Simulator Hardware System

초록

As technologies related to autonomous driving have been actively carried out recently, verification and reliability evaluation of autonomous driving systems are emerging as major issues. Driving simulator technology is also being re-examined to verify development of autonomous driving algorithms equipped with artificial intelligence (AI). Although AI learning data can be generated and secured by acquiring GNSS, LiDAR, RADAR, and V2X data which are essential sensors of existing self-driving systems, there is still a limit to collecting actual data due to the lack of functional support in conducting performance evaluation of multiple vehicles equipped with self-driving systems. In this paper, to secure these shortcomings, a multiple agent server network system for autonomous driving hardware systems is established for the easy test of self-driving algorithms in parallel as well as the efficient use of V2X even in the simulation level. Through communication between the server and several clients, sensor data of self-driving vehicle can be acquired, and based on this, the performance and evaluation of algorithms of self-driving system based on AI algorithm can be enhanced.