Multi-Sensor Information Ensemble-Based Automatic Parking System for Vehicle Parallel/Nonparallel Initial State.

Area of Science:

• Robotics and Autonomous Systems
• Computer Vision
• Control Engineering

Background:

• Automatic parking systems aim for automated vehicle maneuvering into designated spaces.
• Key challenges include accurate parking space recognition and robust trajectory generation, especially for non-parallel initial vehicle states.

Purpose of the Study:

• To propose a novel multi-sensor information ensemble for enhanced parking space recognition.
• To develop an adaptive trajectory generation method that is robust to non-parallel initial vehicle states.
• To improve the overall performance of automatic parking systems.

Main Methods:

• Utilized a multi-sensor information ensemble for improved parking space recognition.
• Developed an adaptive trajectory generation algorithm accounting for non-parallel initial conditions.
• Validated the proposed methods through both simulation and real-world vehicle experiments.

Main Results:

• The multi-sensor ensemble significantly improved parking space recognition accuracy.
• The adaptive trajectory generation method demonstrated robustness for vehicles in non-parallel initial states.
• Experimental results confirmed enhanced automatic parking system performance.

Conclusions:

• The proposed multi-sensor information ensemble and adaptive trajectory generation method effectively address key challenges in automatic parking.
• The approach offers a robust solution for automated parking, even under difficult initial vehicle orientations.