Gear Prediction System Based on Ultrasonic Radar and Machine Learning

Abstract With the rapid development of intelligent vehicles, the emergence of various assisted driving functions has enhanced safety during driving. However, the current mode of shifting in vehicles is mainly through the driver actively operating the gear lever. This mode of shifting not only reduces the usable space inside the vehicle due to the presence of the shifting mechanism but also, because shifting relies more heavily on the driver’s judgment of the environment, incidents of traffic accidents caused by driver’s misoperations and misjudgments occur frequently. To address this issue, this paper proposes a method for predicting the driver’s intention to move forward or backward based on ultrasonic radar. Firstly, a method for detecting obstacles around the vehicle using ultrasonic radar is proposed, which utilizes a pair of ultrasonic radars arranged in an array to locate obstacles around the vehicle based on the principle of elliptical positioning. Based on the location information of obstacles, a simplified vehicle kinematics model is used in conjunction with basic vehicle parameter information to determine the vehicle’s passability. Furthermore, a dataset for different gear shifting targets of the driver in various scenarios is established and trained, equipping the model with the capability to automatically shift gears for the vehicle in different situations. Experimental results show that this method can accurately select the appropriate gear for the vehicle in daily shifting scenarios, thereby enhancing the convenience and safety of driving for the driver in different shifting contexts.