Summary of Pre-trained Vision Models As Perception Backbones For Safety Filters in Autonomous Driving, by Yuxuan Yang and Hussein Sibai
Pre-Trained Vision Models as Perception Backbones for Safety Filters in Autonomous Driving
by Yuxuan Yang, Hussein Sibai
First submitted to arxiv on: 29 Oct 2024
Categories
- Main: Computer Vision and Pattern Recognition (cs.CV)
- Secondary: Machine Learning (cs.LG); Systems and Control (eess.SY)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The paper addresses the challenge of designing safety filters for end-to-end vision-based autonomous driving controllers in high-dimensional settings. To achieve this, it leverages frozen pre-trained vision representation models as perception backbones to design vision-based safety filters. The approach is inspired by these models’ success as backbones of robotic control policies. The offline performance of four common pre-trained vision models is empirically evaluated in this context. Three existing methods are tried for training safety filters for black-box dynamics, as the dynamics over representation spaces are not known. The results show that the proposed filters are competitive with those given ground truth state information. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Autonomous driving has made great progress, but it’s still important to keep drivers safe. One way to do this is by using special filters to control the car and prevent accidents. But designing these filters for autonomous driving can be tricky because there’s a lot of data involved. In this paper, researchers came up with a new approach that uses pre-trained models to design these safety filters. They tested four different types of models and found that they worked well. This could help make self-driving cars safer. |
