Prediction of brake pedal aperture for automatic wheel loader based on Deep LSTM RNN

In the complex and changeable driving environment, the loader's autonomous decision-making and planning ability is seriously insufficient, and it cannot make predictions about upcoming behaviors based on historical experience and current environment like human drivers. In view of this, we carry...

Full description

Saved in:
Bibliographic Details
Published in2023 CAA Symposium on Fault Detection, Supervision and Safety for Technical Processes (SAFEPROCESS) pp. 1 - 6
Main Authors Yong-Kang, Su, Jun-Ren, Shi, Chang-Hao, Piao, Ke-Xin, Li, Ying-Jie, Tang, Yu-Fei, Liang
Format Conference Proceeding
LanguageEnglish
Published IEEE 22.09.2023
Subjects
Apertures
Automatic drive
Brake pedal aperture
Decision making
Deep learning
Planning
Prediction
Support vector machines
Vehicle driving
Wheel loader
Wheels
Online AccessGet full text

Cover

More Information
Summary:In the complex and changeable driving environment, the loader's autonomous decision-making and planning ability is seriously insufficient, and it cannot make predictions about upcoming behaviors based on historical experience and current environment like human drivers. In view of this, we carry out the following research. The research background is the typical operation mode of loader with extremely frequent braking action, and the research core is the prediction of brake pedal aperture for automatic wheel loader. By combining the driving data of experienced drivers in different driving environments with deep learning, a deep long short-term memory network was constructed to predict the brake pedal aperture for different braking types. Finally, the classification necessity and prediction accuracy of the deep LSTM network are verified. The results show that the prediction performance of the complete LSTM network with the classification function is significantly better than that of the imperfect LSTM network without the classification function; The built-in LSTM can be used to assess brake pedal play under various operating conditions, and the predicted value of the brake pedal aperture is very close to the actual value of the skilled driver, indicating that the prediction result conforms to the braking law of the skilled driver. The research method can provide a reference for the prediction of autonomous wheel loader in the decision-making and planning process.
DOI:10.1109/SAFEPROCESS58597.2023.10295592