Cheng Peng

Cheng Peng, Masayoshi Tomizuka

In the autonomous driving area, interaction between vehicles is still a piece of puzzle which has not been fully resolved. The ability to intelligently and safely interact with other vehicles can not only improve self driving quality but also be beneficial to the global driving environment. In this paper, a Bayesian persuasive driving algorithm based on optimization is proposed, where the ego vehicle is the persuader (information sender) and the surrounding vehicle is the persuadee (information receiver). In the persuasion process, the ego vehicle aims at changing the surrounding vehicle's posterior belief of the world state by providing certain information via signaling in order to achieve a lower cost for both players. The information received by the surrounding vehicle and its belief of the world state are described by Gaussian distributions. Simulation results in several common traffic scenarios are provided to demonstrate the proposed algorithm's capability of handling interaction situations involving surrounding vehicles with different driving characteristics.

Liting Sun, Cheng Peng, Wei Zhan et al.

For safe and efficient planning and control in autonomous driving, we need a driving policy which can achieve desirable driving quality in long-term horizon with guaranteed safety and feasibility. Optimization-based approaches, such as Model Predictive Control (MPC), can provide such optimal policies, but their computational complexity is generally unacceptable for real-time implementation. To address this problem, we propose a fast integrated planning and control framework that combines learning- and optimization-based approaches in a two-layer hierarchical structure. The first layer, defined as the "policy layer", is established by a neural network which learns the long-term optimal driving policy generated by MPC. The second layer, called the "execution layer", is a short-term optimization-based controller that tracks the reference trajecotries given by the "policy layer" with guaranteed short-term safety and feasibility. Moreover, with efficient and highly-representative features, a small-size neural network is sufficient in the "policy layer" to handle many complicated driving scenarios. This renders online imitation learning with Dataset Aggregation (DAgger) so that the performance of the "policy layer" can be improved rapidly and continuously online. Several exampled driving scenarios are demonstrated to verify the effectiveness and efficiency of the proposed framework.