GS-Playground: A High-Throughput Photorealistic Simulator for Vision-Informed Robot Learning
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2026
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866915962956021760 |
|---|---|
| author | Jia, Yufei Zhang, Heng Zhang, Ziheng Wu, Junzhe Yu, Mingrui Wang, Zifan Jiang, Dixuan Li, Zheng Cao, Chenyu Yu, Zhuoyuan Yang, Xun Ge, Haizhou Zhang, Yuchi Zhang, Jiayuan Huang, Zhenbiao Liu, Tianle Chen, Shenyu Wang, Jiacheng Xie, Bin Yao, Xuran Deng, Xiwa Wang, Guangyu Zhang, Jinzhi Hao, Lei Chen, Zhixing Chen, Yuxiang Wang, Anqi Tian, Hongyun Yan, Yiyi Cao, Zhanxiang Jiang, Yizhou Shao, Hanyang Li, Yue Shi, Lu Chen, Bokui Sui, Wei Cui, Hanqing Qin, Yusen Huang, Ruqi Han, Lei Wang, Tiancai Zhou, Guyue |
| author_facet | Jia, Yufei Zhang, Heng Zhang, Ziheng Wu, Junzhe Yu, Mingrui Wang, Zifan Jiang, Dixuan Li, Zheng Cao, Chenyu Yu, Zhuoyuan Yang, Xun Ge, Haizhou Zhang, Yuchi Zhang, Jiayuan Huang, Zhenbiao Liu, Tianle Chen, Shenyu Wang, Jiacheng Xie, Bin Yao, Xuran Deng, Xiwa Wang, Guangyu Zhang, Jinzhi Hao, Lei Chen, Zhixing Chen, Yuxiang Wang, Anqi Tian, Hongyun Yan, Yiyi Cao, Zhanxiang Jiang, Yizhou Shao, Hanyang Li, Yue Shi, Lu Chen, Bokui Sui, Wei Cui, Hanqing Qin, Yusen Huang, Ruqi Han, Lei Wang, Tiancai Zhou, Guyue |
| contents | Embodied AI research is undergoing a shift toward vision-centric perceptual paradigms. While massively parallel simulators have catalyzed breakthroughs in proprioception-based locomotion, their potential remains largely untapped for vision-informed tasks due to the prohibitive computational overhead of large-scale photorealistic rendering. Furthermore, the creation of simulation-ready 3D assets heavily relies on labor-intensive manual modeling, while the significant sim-to-real physical gap hinders the transfer of contact-rich manipulation policies. To address these bottlenecks, we propose GS-Playground, a multi-modal simulation framework designed to accelerate end-to-end perceptual learning. We develop a novel high-performance parallel physics engine, specifically designed to integrate with a batch 3D Gaussian Splatting (3DGS) rendering pipeline to ensure high-fidelity synchronization. Our system achieves a breakthrough throughput of 10^4 FPS at 640x480 resolution, significantly lowering the barrier for large-scale visual RL. Additionally, we introduce an automated Real2Sim workflow that reconstructs photorealistic, physically consistent, and memory-efficient environments, streamlining the generation of complex simulation-ready scenes. Extensive experiments on locomotion, navigation, and manipulation demonstrate that GS-Playground effectively bridges the perceptual and physical gaps across diverse embodied tasks. Project homepage: https://gsplayground.github.io. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_25459 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | GS-Playground: A High-Throughput Photorealistic Simulator for Vision-Informed Robot Learning Jia, Yufei Zhang, Heng Zhang, Ziheng Wu, Junzhe Yu, Mingrui Wang, Zifan Jiang, Dixuan Li, Zheng Cao, Chenyu Yu, Zhuoyuan Yang, Xun Ge, Haizhou Zhang, Yuchi Zhang, Jiayuan Huang, Zhenbiao Liu, Tianle Chen, Shenyu Wang, Jiacheng Xie, Bin Yao, Xuran Deng, Xiwa Wang, Guangyu Zhang, Jinzhi Hao, Lei Chen, Zhixing Chen, Yuxiang Wang, Anqi Tian, Hongyun Yan, Yiyi Cao, Zhanxiang Jiang, Yizhou Shao, Hanyang Li, Yue Shi, Lu Chen, Bokui Sui, Wei Cui, Hanqing Qin, Yusen Huang, Ruqi Han, Lei Wang, Tiancai Zhou, Guyue Robotics 68T40 I.2.9 Embodied AI research is undergoing a shift toward vision-centric perceptual paradigms. While massively parallel simulators have catalyzed breakthroughs in proprioception-based locomotion, their potential remains largely untapped for vision-informed tasks due to the prohibitive computational overhead of large-scale photorealistic rendering. Furthermore, the creation of simulation-ready 3D assets heavily relies on labor-intensive manual modeling, while the significant sim-to-real physical gap hinders the transfer of contact-rich manipulation policies. To address these bottlenecks, we propose GS-Playground, a multi-modal simulation framework designed to accelerate end-to-end perceptual learning. We develop a novel high-performance parallel physics engine, specifically designed to integrate with a batch 3D Gaussian Splatting (3DGS) rendering pipeline to ensure high-fidelity synchronization. Our system achieves a breakthrough throughput of 10^4 FPS at 640x480 resolution, significantly lowering the barrier for large-scale visual RL. Additionally, we introduce an automated Real2Sim workflow that reconstructs photorealistic, physically consistent, and memory-efficient environments, streamlining the generation of complex simulation-ready scenes. Extensive experiments on locomotion, navigation, and manipulation demonstrate that GS-Playground effectively bridges the perceptual and physical gaps across diverse embodied tasks. Project homepage: https://gsplayground.github.io. |
| title | GS-Playground: A High-Throughput Photorealistic Simulator for Vision-Informed Robot Learning |
| topic | Robotics 68T40 I.2.9 |
| url | https://arxiv.org/abs/2604.25459 |