Abstract: Against the backdrop of the continuous increase in the density of urban motor vehicles and the growing demand for parking in complex environments, in order to solve the problems such as large detection blind areas and delayed dynamic responses caused by the traditional automatic parking system relying on a single sensor, this study aims to design a low-cost automatic parking system based on multi-sensor fusion to achieve precise parking in the environment without parking lines. The system takes STM32F103C8T6 as the core controller, realizes the detection of obstacles around the vehicle body through four TCRT5000 infrared emission sensors, and combines HC-SR04 ultrasonic sensors to accurately obtain the distance information at the rear of the vehicle. The small car can achieve reversing into a parking space or lateral parking without a stop line. It adopts a multi-sensor data fusion algorithm and generates the optimal parking path based on geometric constraints. Experiments show that in the simulated parking space, the success rate of the system reversing into the parking space is 96%, with the entire process taking 5.2 s. The success rate of side parking is 93%, with the entire process taking 6.5 s. This study provides a feasible technical path for automatic parking without parking lines.

Key words: STM32 microcontroller; automatic parking system; infrared reflection sensor; ultrasonic sensor; without parking lines

摘要： 在城市机动车密度持续攀升、复杂环境泊车需求日益增长的背景下，为解决传统自动 泊车系统依赖单一传感器导致的检测盲区大、动态响应迟滞等问题，该研究旨在设计一种基 于多传 感器融合的低成本 自动泊车系统，实 现无停车线环境下 的精准泊车。系统以 STM32F103C8T6 单片机为核心控制器，通过四个 TCRT5000 红外发射传感器实现车身周围障 碍物检测，结合 HC-SR04 超声波传感器精准获取车尾距离信息。小车可以实现无停车线情况 下的倒车入库或者侧方停车，采用多传感器数据融合算法并基于几何约束生成最优泊车路径。 实验表明，在模拟车位中，系统倒车入库成功率 96%，全程耗时 5.2 s，侧方停车成功率 93%， 全程耗时 6.5 s。研究为无停车线时的自动泊车提供了可行的技术路径。

关键词: STM32 单片机；自动泊车系统；红外反射传感器；超声波传感器；无停车线