ROCVNov 13, 2020

Benchmarking Domain Randomisation for Visual Sim-to-Real Transfer

arXiv:2011.07112v327 citations
AI Analysis

This work provides practical guidelines for robotics researchers using domain randomisation, but it is incremental as it benchmarks existing methods rather than introducing new ones.

The paper benchmarks design choices in domain randomisation for visual sim-to-real transfer, finding that a small number of high-quality images outperforms many low-quality ones and that both distractors and textures are crucial for generalisation to new environments.

Domain randomisation is a very popular method for visual sim-to-real transfer in robotics, due to its simplicity and ability to achieve transfer without any real-world images at all. Nonetheless, a number of design choices must be made to achieve optimal transfer. In this paper, we perform a comprehensive benchmarking study on these different choices, with two key experiments evaluated on a real-world object pose estimation task. First, we study the rendering quality, and find that a small number of high-quality images is superior to a large number of low-quality images. Second, we study the type of randomisation, and find that both distractors and textures are important for generalisation to novel environments.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes