Summary of Diffusion-es: Gradient-free Planning with Diffusion For Autonomous Driving and Zero-shot Instruction Following, by Brian Yang et al.
Diffusion-ES: Gradient-free Planning with Diffusion for Autonomous Driving and Zero-Shot Instruction Following
by Brian Yang, Huangyuan Su, Nikolaos Gkanatsios, Tsung-Wei Ke, Ayush Jain, Jeff Schneider, Katerina Fragkiadaki
First submitted to arxiv on: 9 Feb 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Robotics (cs.RO)
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 This paper proposes a new method called DiffusionES that combines gradient-free optimization with trajectory denoising to optimize black-box non-differentiable objectives while staying in the data manifold. Building on previous work, such as reward-gradient guided denoising, DiffusionES samples trajectories from a diffusion model and scores them using a black-box reward function. It then mutates high-scoring trajectories using a truncated diffusion process that applies a small number of noising and denoising steps, allowing for efficient exploration of the solution space. The authors demonstrate state-of-the-art performance on nuPlan, an established closed-loop planning benchmark for autonomous driving, outperforming existing sampling-based planners, reactive deterministic or diffusion-based policies, and reward-gradient guidance. Additionally, they show that DiffusionES can optimize non-differentiable language-shaped reward functions generated by few-shot LLM prompting. When guided by a human teacher, the method can generate novel, highly complex behaviors not present in the training data. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper is about a new way to make decisions and control systems using computers. It’s called DiffusionES, and it helps machines learn from their experiences and adapt to changing situations. The authors wanted to solve a specific problem called closed-loop planning for autonomous driving. They tested their method on a benchmark and found that it worked better than existing methods. This new approach can even understand non-technical language and follow instructions from humans. It’s an important step forward in making machines more intelligent and able to work with people. |
Keywords
* Artificial intelligence * Diffusion * Diffusion model * Few shot * Optimization * Prompting
