Lane Change Classification and Prediction with Action Recognition Networks

TL;DR

This paper introduces an end-to-end action recognition framework using RGB video data to classify and predict lane change intentions in autonomous driving, outperforming previous methods that rely on physical variables.

Contribution

The work presents a novel application of action recognition models to lane change prediction using only camera data, eliminating the need for pre-processed physical variables.

Findings

Achieved state-of-the-art lane change classification accuracy with RGB videos.

Demonstrated that action recognition models effectively extract lane change motions.

Proposed a new method to enhance motion feature extraction from video data.

Abstract

Anticipating lane change intentions of surrounding vehicles is crucial for efficient and safe driving decision making in an autonomous driving system. Previous works often adopt physical variables such as driving speed, acceleration and so forth for lane change classification. However, physical variables do not contain semantic information. Although 3D CNNs have been developing rapidly, the number of methods utilising action recognition models and appearance feature for lane change recognition is low, and they all require additional information to pre-process data. In this work, we propose an end-to-end framework including two action recognition methods for lane change recognition, using video data collected by cameras. Our method achieves the best lane change classification results using only the RGB video data of the PREVENTION dataset. Class activation maps demonstrate that action recognition models can efficiently extract lane change motions. A method to better extract motion clues is also proposed in this paper.