Authors: Matteo Cederle, Marco Fabris, Gian Antonio Susto

Published on: May 14, 2024

Impact Score: 7.2

Arxiv code: Arxiv:2405.08655

Summary

• What is new: Introduction of a distributed multi-agent reinforcement learning approach and prioritized scenario replay for autonomous intersection management.
• Why this is important: High cost and complexity of centralized servers for managing autonomous vehicle intersections.
• What the research proposes: A novel multi-agent reinforcement learning algorithm utilizing 3D surround view technology, eliminating the need for centralised control.
• Results: Superior performance in virtual environment tests on the SMARTS platform, surpassing conventional benchmarks.

Technical Details

Technological frameworks used: SMARTS platform

Models used: Multi-agent reinforcement learning (MARL)

Data used: 3D surround view technology data

Potential Impact

Automotive industry, particularly companies developing or employing vehicle intersection control systems; Traffic management technology providers.

Want to implement this idea in a business?

We have generated a startup concept here: IntersectAI.