Fine-Tuning the TOPSIS Technique and Transferring Knowledge of Different Driving Styles

This study investigates trajectory generation methods for automated driving by refining existing approaches using polynomial algorithms and TOPSIS. While initial heuristics adjusted vectors, alternative methods such as Bayesian optimization and experience transfer were explored. Bayesian optimization initially achieves higher efficiency and comfort scores (0.76 and 0.45), but experience transfer allows for adaptive driving styles, balancing metrics (efficiency 0.51-0.78, safety 0.38-0.56, comfort 0.22-0.38, LCFQI 0.36-0.64) according to driving style and static obstacles. Experience Transfer provides immediacy by leveraging prior knowledge without exhaustive search, making it a favorable choice for rapid, critical implementations in complex driving scenarios.