Research map / Reinforcement learning / Visual and representation RL
Tac2Real: Reliable and GPU Visuotactile Simulation for Online Reinforcement Learning and Zero-shot Real-World Deployment
Ningyu YAN, Shuai Wang, Xing Shen, Hui Wang, Hanqing Wang, Yang Xiang, Jiangmiao Pang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use PNCG-IPC (Tac2Real) tactile simulation instead of penalty-based TacSL for real-world transfer | Tac2Real · Peg Insertion real world | success rate: 0.150 -> 0.917 | Improved |
| Use Tac2Real vs TacSL in simulation | Tac2Real · Peg Insertion simulation | success rate: 0.789 (TacSL) vs 0.776 | No clear effect |
| Use Tac2Real vs Tacchi (MPM) in simulation | Tac2Real · Peg Insertion simulation | success rate: 0.173 -> 0.776 | Improved |
| Add tactile observations | Tac2Real · Nut Threading simulation | success rate: 0.313 -> 0.702 | Improved |
| Add control alignment (level 1) to TacAlign levels 2,3,4 (ablation) | TacAlign · Peg Insertion real world | success rate: 0.533 -> 0.917 | Improved |
| Add task-based calibration (level 3) to TacAlign levels 1,2,4 (ablation) | TacAlign · Peg Insertion real world | success rate: 0.250 -> 0.917 | Improved |
| Add domain randomization (level 4) to TacAlign levels 1,2,3 (ablation) | TacAlign · Peg Insertion real world | success rate: 0.767 -> 0.917 | Improved |
| Optimize controller gains by trajectory alignment (ablation) | TacAlign control alignment · Controller alignment | translation trajectory discrepancy: 11.11 mm -> 2.521 mm | Improved |
| Use Tac2Real instead of TacSL for throughput | Tac2Real · PNCG-IPC parallel performance | FPS | Hurt |
Closest papers by meaning
- Taccel: Scaling Up Vision-based Tactile Robotics via High-performance GPU Simulation (NeurIPS 2025 · similarity 0.90)
- Enhancing Tactile-based Reinforcement Learning for Robotic Control (NeurIPS 2025 · similarity 0.87)
- DIFFTACTILE: A Physics-based Differentiable Tactile Simulator for Contact-rich Robotic Manipulation (ICLR 2024 · similarity 0.87)
- PhysTacGen: Physics-Aware Visual-Tactile Sensor Image Generation (NeurIPS 2026 · similarity 0.86)
- Touch in the Wild: Learning Fine-Grained Manipulation with a Portable Visuo-Tactile Gripper (NeurIPS 2025 · similarity 0.86)
- Multi-fingered Hand Grasps with Visuo-Tactile Fusion via Multi-Agent Deep Reinforcement Learning (AAAI 2025 · similarity 0.85)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
