Summary

SurRoL is an open-source platform designed for surgical robot learning, focusing on generalized task autonomy in laparoscopic procedures. It targets researchers and engineers in surgical robotics, offering a reinforcement learning-centric environment compatible with the da Vinci Research Kit (dVRK). The platform aims to bridge the simulation-to-real gap, enabling advanced robotic control and automation.

How It Works

The system leverages PyBullet and Taichi for physics simulation, providing an OpenAI Gym-style API for reinforcement learning agents. It integrates with the dVRK for real-world robot control and can incorporate human interaction via haptic devices. The core VPPV framework manages training data generation, state regression for perceptual models, and RL policy learning. Deployment is supported on dVRK robots and through the Sentire system for ex vivo/in vivo experiments, facilitating a comprehensive research workflow.

Quick Start & Requirements

• Primary install/run command: No single command; involves running Python scripts for data generation, state regression training, RL policy training, and deployment.
• Prerequisites: Ubuntu 20.04, Python 3.7, ROS Noetic, dVRK 2.1. Requires following specific guides for dVRK hardware/software setup, camera calibration, and hand-eye calibration.
• Links: Project Website, IROS'21 Paper, IROS'21 Code.

Highlighted Details

• Full dVRK compatibility for real-world robotic surgery simulation and control.
• OpenAI Gym-style API simplifies integration with standard RL libraries.
• Zero-shot sim-to-real transfer capabilities aim to reduce real-world training overhead.
• Includes modules for data-driven scene simulation and intelligent haptic guidance.
• Supports deployment on both game-based training tasks and advanced experimental setups (Sentire).

Maintenance & Community

The project is associated with multiple high-impact publications from 2021-2025, indicating active research development. Specific community channels (e.g., Discord, Slack) or dedicated maintainer information are not detailed in the provided README.

Licensing & Compatibility

Released under the permissive MIT license, allowing for broad use, modification, and distribution, including in commercial applications.

Limitations & Caveats

The platform's setup is tightly coupled with specific versions of ROS (Noetic) and dVRK (2.1), potentially requiring significant effort for integration with newer or different robotic systems. The research-oriented nature implies that stability and ease of use for non-expert users may be secondary to experimental flexibility.