Authors: Daniel Bogdoll, Jing Qin, Moritz Nekolla, Ahmed Abouelazm, Tim Joseph, J. Marius Zöllner
Published on: February 06, 2024
Impact Score: 8.38
Arxiv code: Arxiv:2402.04168
Summary
- What is new: The introduction of Informed Reinforcement Learning that integrates a structured rulebook as a knowledge source in autonomous driving scenarios.
- Why this is important: The simplicity of scenarios in autonomous driving research and the usage of non-interpretable control commands and unstructured reward designs.
- What the research proposes: Using Informed Reinforcement Learning to incorporate a structured rulebook for learning trajectories with a dynamic, situation-aware reward design.
- Results: High completion rates in complex scenarios using recent model-based agents.
Technical Details
Technological frameworks used: Informed Reinforcement Learning
Models used: Recent model-based agents
Data used: Trajectories in autonomous driving scenarios
Potential Impact
Autonomous driving industry, companies developing self-driving cars, and regulatory bodies involved in traffic management and safety.
Want to implement this idea in a business?
We have generated a startup concept here: SafePath AI.
Leave a Reply