Study on Design of Obstacle Collision Prevention System using LiDAR based on Multi-body Dynamics Simulator

초록

This study aimed to develop a simulator to design an obstacle collision prevention system using LiDAR, optimization of vehicle control performance and safety based on multi-body dynamics simulation using Simulink. The simulator incorporates vehicle dynamics theory and includes a LiDAR model for obstacle detection, using PID control and Pure-pursuit control algorithms for longitudinal and lateral control. Obstacle avoidance was managed by adjusting the PID gain and look ahead distance based on LiDAR measurements. Various driving scenarios were analyzed in simulation for performance, stability, and safety to adapt in AI model. Initial results showed reduced tracking performance and stability when obstacles were present, and obstacle prevention system operated. To optimize controller parameters, a feedforward neural network and genetic algorithm were used, enhancing vehicle control accuracy and collision avoidance by 11% and 29%, respectively, but decreasing rollover stability by 23%. The results showed that the design and optimization of obstacle collision avoidance system and control performance can be progressed in virtual simulator environment.