Research on Automatic Parking System Based on Multi-information Fusion

Chuan-wei Zhang ； Hong-jun Zeng ； Meng-yue Yang ； Jian-ping Wen ； Man-zhi Yang

《電腦學刊》 31卷5期 (2020/10) Pp. 14-29

https://doi.org/10.3966/199115992020103105002

摘要

Aiming at the requirement of automatic parking of intelligent vehicles in complex environment, this paper proposes an automatic parking method based on multi-information fusion. The method can not only improve the accuracy of the intelligent recognition of the parking space, but also ensure the safety and efficiency in the path planning process. The automatic parking system first uses multiple sensors to identify the relevant feature information of the surrounding environment of the parking space, and then calculates the output parking space recognition result according to the fuzzy inference system (FIS) and the two-way breadth-first search algorithm based on the rasterized map idea is used to carry out the corresponding path planning. Finally, the simulation analysis of the complex environment is carried out. The simulation results verify the rationality and effectiveness of the intelligent identification and parking path planning scheme. Then the self-developed automatic parking system was built in the experimental car for real test. The test results show that the correct rate of the parking system for automatic parking in complex environments is 89%-98%, with an average of 93.75%. The collision avoidance rate during the path planning process is 96% to 99%, and the average value is 98%, which verifies the good effect of the automatic parking method.

關鍵字

automatic parking ； multi-information fusion ； parking space identification ； path planning