Reproduce “How Simulation Helps Autonomous Driving:A Survey of Sim2real, Digital Twins, and Parallel Intelligence” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Xuemin Hu, Shen Li, Tingyu Huang, Bo Tang, Rouxing Huai, Long Chen
The authors survey ways to address the reality gap between simulated and real autonomous driving. They organize the literature into sim2real transfer, digital twins that connect virtual counterparts with physical entities, and parallel intelligence that uses computational experiments and feedback. The survey offers a conceptual map, not a controlled comparison, so it does not establish which approach is most effective or that any closes the gap.
Key insights
- Sim2real, Digital Twins, and Parallel Intelligence represent an evolving trajectory towards increasingly integrated and intelligent virtual-real systems for robust autonomous driving.
- Each paradigm offers distinct advantages: sim2real focuses on adapting simulated knowledge, DTs enable real-time virtual modeling, and PI facilitates dynamic interaction for complex system management.
- Bridging the reality gap effectively requires addressing challenges related to model generalizability, data scarcity, efficient multi-modal data analysis, and the development of sophisticated simulation capabilities.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.