Authors: Shu Li, Jingxuan Yang, Honglin He, Yi Zhang, Jianming Hu, Shuo Feng

Published on: February 02, 2024

Impact Score: 8.12

Arxiv code: Arxiv:2402.01795

Summary

• What is new: Introduces the concept of ‘few-shot testing’ (FST) for autonomus vehicles (AVs) and proposes a new framework for it.
• Why this is important: Testing AVs within a very limited number of attempts poses significant challenges and uncertainties in safety evaluation.
• What the research proposes: A systematic FST framework that optimizes testing scenario utilization and minimizes uncertainties by leveraging surrogate models and neighborhood coverage.
• Results: Demonstrated a significant reduction in testing error and variance in cut-in scenarios, especially under strict testing limitations.

Technical Details

Technological frameworks used: Few-shot testing (FST)

Models used: Surrogate models (SMs)

Data used: Prior information from surrogate models

Potential Impact

Automotive industry, autonomous vehicle manufacturers, safety testing services

Want to implement this idea in a business?

We have generated a startup concept here: OptiTest AV.