Authors: Haicheng Liao, Shangqian Liu, Yongkang Li, Zhenning Li, Chengyue Wang, Bonan Wang, Yanchen Guan, Chengzhong Xu

Published on: February 06, 2024

Impact Score: 8.45

Arxiv code: Arxiv:2402.04318

Summary

• What is new: Introduction of an ‘adaptive visual sector’ mimicking human attention allocation and a ‘dynamic traffic graph’ for trajectory prediction in AVs.
• Why this is important: Autonomous vehicles face challenges in trajectory prediction in mixed autonomy environments.
• What the research proposes: An interdisciplinary model that integrates human cognition principles with Convolutional Neural Networks (CNN) and Graph Attention Networks (GAT).
• Results: Model outperforms existing baselines by at least 15.2%, 19.4%, and 12.0% on NGSIM, HighD, and MoCAD datasets, respectively.

Technical Details

Technological frameworks used: CNN, GAT

Models used: GAVA model

Data used: NGSIM, HighD, MoCAD datasets

Potential Impact

Automotive and technology companies developing autonomous vehicle solutions could greatly benefit or face disruption.

Want to implement this idea in a business?

We have generated a startup concept here: CogniPath.