PPAD: Iterative Interactions of Prediction and Planning for End-to-end Autonomous Driving

Zhili Chen,Maosheng Ye,Shuangjie Xu,Tongyi Cao,Qifeng Chen

2024-07-22

Abstract:We present a new interaction mechanism of prediction and planning for end-to-end autonomous driving, called PPAD (Iterative Interaction of Prediction and Planning Autonomous Driving), which considers the timestep-wise interaction to better integrate prediction and planning. An ego vehicle performs motion planning at each timestep based on the trajectory prediction of surrounding agents (e.g., vehicles and pedestrians) and its local road conditions. Unlike existing end-to-end autonomous driving frameworks, PPAD models the interactions among ego, agents, and the dynamic environment in an autoregressive manner by interleaving the Prediction and Planning processes at every timestep, instead of a single sequential process of prediction followed by planning. Specifically, we design ego-to-agent, ego-to-map, and ego-to-BEV interaction mechanisms with hierarchical dynamic key objects attention to better model the interactions. The experiments on the nuScenes benchmark show that our approach outperforms state-of-the-art methods.

Computer Vision and Pattern Recognition,Robotics

What problem does this paper attempt to address?

This paper attempts to address the issue of how to better integrate the interaction between prediction and planning modules in end-to-end autonomous driving systems. Specifically:

1. **Problems with traditional methods**: Traditional autonomous driving systems are usually divided into multiple modules (such as localization, perception, tracking, prediction, planning, and control). This modular approach can lead to the accumulation of errors across modules, affecting the overall performance of the system.

2. **Advantages and shortcomings of end-to-end methods**: End-to-end methods simplify system design, but existing methods often consider only one-time prediction and planning steps, failing to fully simulate the dynamic interaction process of traffic participants.

3. **Proposed new method**: The paper proposes a new iterative interaction mechanism for prediction and planning (PPAD). This mechanism better simulates the complex interactions between traffic participants by interleaving the prediction and planning processes at each time step. This approach not only improves prediction accuracy but also makes planning more reasonable and safer.

In summary, PPAD aims to enhance the decision-making ability and safety of autonomous vehicles in complex dynamic environments by improving the interaction between prediction and planning.