摘要
针对自动泊车系统对不同类型停车位进行识别的要求,提出一种基于距离和视觉信息融合的停车位智能识别方法。建立了根据超声波传感器、视觉信息传感器、里程计信息来辨识停车位两侧车辆姿态和车位参数的模型,根据Mamdani模糊推理计算输出相应的泊车车位识别结果。利用Matlab/Simulink搭建车位自动辨识以及相应的泊车路径仿真模型,针对五个典型的停车位场景进行了仿真分析,仿真结果验证了车位智能识别和泊车路径规划方案的合理性和有效性。将自主开发的自动泊车系统搭载于某轿车进行试验,试验结果表明,该泊车系统对规则和不规则车位的自动识别正确率为88%~98%,平均值为92.8%,从而验证了车位智能识别方法的良好效果。
Aiming at the requirement of the automatic parking system intelligently identifying different types of parking spaces,this paper proposes an recognition method of parking spaces based on distance and visual information fusion. Establishing a model for identifying the vehicle attitude and parking parameters according to the ultrasonic sensors, visual information sensors, odometer information, the controller uses Mamdani fuzzy inference to automatically output the parking plan suitable. Establishing the automatic parking space identification model and parking path simulation model based on Matlab/Simulink, five typical parking spaces are simulated and analyzed, simulation results validated the rationality and effectiveness of intelligent parking identification and parking path planning. Experimenting on a car equipped with this automatic parking system, the results show that the accuracy of the recognition method is 88%-98% for regular and irregular parking spaces, statistical average is 92.8%, which proved the good effect of intelligent recognition of the parking technology. It shows that intelligent identification technology of automatic parking system based on Information Fusion designed has high performance.
出处
《机械工程学报》
EI
CAS
CSCD
北大核心
2017年第22期125-133,共9页
Journal of Mechanical Engineering
基金
国家自然科学基金重点(U1564201)
江苏高校自然学科研究重大项目(16KJA580001)
江苏省产学研前瞻性联合创新(BY2012173)资助项目
关键词
多传感器信息融合
模糊推理
不规则停车位
智能识别
multi-sensor information
fusion
fuzzy reasoning
irregular parkingspaces
intelligent recognition
