What is it about?
This survey explores how physics simulators can help robots learn navigation and manipulation safely and efficiently before being deployed in the real world. It reviews different simulation platforms, their capabilities, hardware requirements, and the challenges of transferring skills from simulation to real robots. The survey also provides practical resources, including datasets, evaluation metrics, simulation platforms, and methods, to help researchers choose suitable simulation tools for different robotic tasks.
Featured Image
Photo by julien Tromeur on Unsplash
Why is it important?
As robots increasingly learn complex skills through simulation, choosing the right simulator has become critical for reliable real-world deployment. However, existing surveys often focus on individual simulation platforms or learning methods without systematically examining how simulator properties affect navigation, manipulation, and sim-to-real transfer. Our survey provides a unified view of these factors, including physics, sensing, rendering, hardware requirements, and task-specific needs. By bringing together recent methods, benchmarks, and practical resources, this work helps researchers better understand simulation limitations, select appropriate tools, and design more reliable robotic systems for real-world applications.
Perspectives
From my perspective, this survey highlights an important challenge for the next generation of embodied AI: simulation should be treated not only as a tool for training robots, but also as a means of understanding and reducing the gap between simulated and real-world behavior. I hope this work can help researchers make more informed choices when selecting simulators and designing experiments, while encouraging closer integration between simulation and real-world robotic systems.
Dr. Dr. Xueyang Kang
University of Melbourne
Read the Original
This page is a summary of: A Survey of Robotic Navigation and Manipulation with Physics Simulators in the Era of Embodied AI, ACM Computing Surveys, October 2026, ACM (Association for Computing Machinery),
DOI: 10.1145/3856802.
You can read the full text:
Resources
Contributors
The following have contributed to this page
