TrajectoryNAS: A Neural Architecture Search for Trajectory Prediction

TL;DR

TrajectoryNAS systematically optimizes the architecture of an end-to-end trajectory prediction algorithm, incorporating all stacked components that are prerequisites for trajectory prediction, including object detection and object tracking, using metaheuristic algorithms.

Abstract

Autonomous driving systems are a rapidly evolving technology that enables driverless car production. Trajectory prediction is a critical component of autonomous driving systems, enabling cars to anticipate the movements of surrounding objects for safe navigation. Trajectory prediction using Lidar point-cloud data performs better than 2D images due to providing 3D information. However, processing point-cloud data is more complicated and time-consuming than 2D images. Hence, state-of-the-art 3D trajectory predictions using point-cloud data suffer from slow and erroneous predictions. This paper introduces TrajectoryNAS, a pioneering method that focuses on utilizing point cloud data for trajectory prediction. By leveraging Neural Architecture Search (NAS), TrajectoryNAS automates the design of trajectory prediction models, encompassing object detection, tracking, and forecasting in a cohesive manner. This approach not only addresses the complex interdependencies among these tasks but also emphasizes the importance of accuracy and efficiency in trajectory modeling. Through empirical studies, TrajectoryNAS demonstrates its effectiveness in enhancing the performance of autonomous driving systems, marking a significant advancement in the field.Experimental results reveal that TrajcetoryNAS yield a minimum of 4.8 higger accuracy and 1.1* lower latency over competing methods on the NuScenes dataset.

Figures (4)

• Figure 1: (Top Row) Cascade methods that independently address detection, tracking, and forecasting, they inherently carry the risk of compounding errors throughout the pipeline. This originates from each sub-module's assumption of receiving perfect input, which rarely holds true in real-world applications. Consequently, errors introduced in earlier stages propagate and magnify downstream, potentially leading to inaccurate final outcomes. (Bottom Row) End-to-end methods that forecast future movement directly from raw data, enabling end-to-end training and benefiting from the joint optimization of object detection, tracking, and prediction tasks.
• Figure 2: TrajectoryNAS state diagram. A model generated from the search space. The generated model trains using the mini dataset. The results are sent back to search space to generate a new model. The best final model is fully trained using the original dataset.
• Figure 3: The overview of TrajcetoryNAS process.
• Figure 4: TrajectoryNAS optimization curve.