Detection of Emergency Braking Intention Based on Pedal-Switching Process

The driver’s reaction time and the brake actuator response time during emergency braking play a crucial role in braking distance. Detecting the driver’s braking intention and reducing actuator reaction time can enhance the safety advantages of the intelligent braking system. This study introduces a braking intent recognition method based on detecting the driver’s transition between the accelerator and brake pedals. This approach can seamlessly integrate with the brake-by-wire (BBW) system, leveraging the advantages of complete decoupling between the brake pedal and the actuator. This study initially examines the braking process and then presents the proposed braking intent recognition technique. The driver’s pedal-switching action is characterized by a time-varying vector, which is integrated into the detection module. Various feature extraction methods are introduced to analyze the accelerator release speed and time-varying vectors under emergency and normal braking scenarios separately. Leveraging these features, a predictive model is developed using machine learning methods. The effectiveness of the proposed detection system and predictive model is validated through pseudo-online simulations. Results demonstrate a high recognition accuracy of 96.2% for the emergency braking intention detection system. Moreover, the proposed method shows the capability to predict braking intentions about 36 ms prior to the brake pedal being pressed during emergency braking and about 589 ms in advance during normal braking. The proposed approach has the potential to reduce braking distances significantly, serving as a valuable enhancement to Active Driving Assistance Systems (ADAS) for improved braking safety.