Imitating Driver Behavior for Fast Overtaking through Bagging Gaussian Process Regression

For tackling with the autonomous driving tasks such as fast overtaking, traditional rule-based methods are inflexible and lack abilities of adapting to rapidly changing environments. However, human drivers performed well and have accumulated a large amount of driving behavior data which can be used as a reliable basis for learning intelligent driving behavior in these environments. The autonomous vehicle can gradually learn to obtain human-like behavior policies for complex driving tasks through imitation learning (IL). In this paper, a Bagging Gaussian Process Regression (Bagging GPR) method is proposed to imitate driver behaviors for dynamic complex driving tasks, which combined the Bagging with the Gaussian Process Regression (GPR). Human driver data are collected through the simulation platform Prescan, and different driver behavior models are trained and verified by collected data. Experiments show that compared with Multi-layer BP, ELM, Regression Tree Ensemble, SVR and GPR, the proposed method can further reduce state-action imitating errors and achieve more robust performance of the driver behavior learning.

[1]  Leo Breiman,et al.  Bagging Predictors , 1996, Machine Learning.

[2]  Mykel J. Kochenderfer,et al.  Imitating driver behavior with generative adversarial networks , 2017, 2017 IEEE Intelligent Vehicles Symposium (IV).

[3]  William Whittaker,et al.  Autonomous driving in urban environments: Boss and the Urban Challenge , 2008, J. Field Robotics.

[4]  Julius Ziegler,et al.  Making Bertha Drive—An Autonomous Journey on a Historic Route , 2014, IEEE Intelligent Transportation Systems Magazine.

[5]  Carl E. Rasmussen,et al.  Gaussian processes for machine learning , 2005, Adaptive computation and machine learning.

[6]  Luke Fletcher,et al.  A perception-driven autonomous urban vehicle , 2008 .

[7]  Tao Chen,et al.  Bagging for Gaussian process regression , 2009, Neurocomputing.

[8]  Jianxiong Xiao,et al.  DeepDriving: Learning Affordance for Direct Perception in Autonomous Driving , 2015, 2015 IEEE International Conference on Computer Vision (ICCV).

[9]  Alexey Dosovitskiy,et al.  End-to-End Driving Via Conditional Imitation Learning , 2017, 2018 IEEE International Conference on Robotics and Automation (ICRA).

[10]  Peter King,et al.  Odin: Team VictorTango's entry in the DARPA Urban Challenge , 2008, J. Field Robotics.

[11]  Dean Pomerleau,et al.  Efficient Training of Artificial Neural Networks for Autonomous Navigation , 1991, Neural Computation.

[12]  Geoffrey J. Gordon,et al.  No-Regret Reductions for Imitation Learning and Structured Prediction , 2010, ArXiv.

[13]  Lawrence D. Jackel,et al.  Explaining How a Deep Neural Network Trained with End-to-End Learning Steers a Car , 2017, ArXiv.

[14]  Yang Gao,et al.  End-to-End Learning of Driving Models from Large-Scale Video Datasets , 2016, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[15]  Hani S. Mahmassani,et al.  Modeling Lane-Changing Behavior in a Connected Environment: A Game Theory Approach , 2015 .

[16]  Junqiang Xi,et al.  Learning Driver-Specific Behavior for Overtaking: A Combined Learning Framework , 2018, IEEE Transactions on Vehicular Technology.