RCM-Constrained MIS Demos (Franka Panda + MuJoCo)

This repository contains three small Python demos for my thesis on Remote Center of Motion (RCM) control for minimally invasive surgery (MIS) using a Franka Panda model in MuJoCo.
The goal is to keep a fixed trocar point while the instrument tip tracks a desired trajectory inside the patient.

---

Methods implemented

Script	Control level	Brief description	Model XML
src/paper1.py	Kinematic (velocity)	Task-space resolved-rate control with an RCM constraint as in Aghakhani et al. (ICRA 2013).	models/panda.xml
src/paper2.py	Dynamic (torque)	Redundant torque-controlled formulation with an RCM constraint as in Sandoval Arévalo et al. (IROS 2017).	models/panda1.xml
src/paper2_contact.py	Dynamic (torque) + contact	Same as paper2, but adds physical contact between the instrument and the trocar.	models/panda2.xml

Each script loads its model with a relative path (no hard-coded desktop paths), so it should work on any machine after cloning.

---

Repository Structure

• src/ – Python scripts
• models/ – panda.xml, panda1.xml, panda2.xml and the assets/ they need
• models/LICENSE – License for the Panda model
• requirements.txt – Python dependencies
• .gitignore – Files Git should ignore

Scripts and Models

• src/paper1.py → uses models/panda.xml (Paper 1 baseline)
• src/paper2.py → uses models/panda1.xml (Paper 2 baseline)
• src/paper2_contact.py → uses models/panda2.xml (Paper 2 with extra/contact features)

Note: Each script loads its model with a relative path so it runs on any computer.

Prerequisites

• Python 3.10+
• MuJoCo installed on your machine (per MuJoCo’s install instructions)

Setup

pip install -r requirements.txt

How to run

From the repository root:

# Kinematic RCM (Aghakhani13)
python src/paper1.py

# Torque-level RCM (Sandoval17)
python src/paper2.py

# Torque-level RCM + trocar contact
python src/paper2_contact.py

Each demo opens the MuJoCo viewer and drives the instrument tip along a preset trajectory while enforcing the RCM at the trocar.

If the viewer window fails to open, install system OpenGL/GLFW libs (e.g. Ubuntu: sudo apt-get install -y libglfw3 libgl1).

Configuration notes

The desired tip trajectory and controller gains are defined near the top of each script.

XML variants differ only where needed (e.g., contact sites, tool attachments, actuator type and limits). Keep models/assets/ next to the XMLs.

Scripts load models with a relative path (portable):

from pathlib import Path
HERE = Path(__file__).resolve().parent
MODEL_XML = HERE.parent / "models" / "<panda.xml | panda1.xml | panda2.xml>"

References

N. Aghakhani, M. Geravand, N. Shahriari, M. Vendittelli, and G. Oriolo, “Task control with remote center of motion constraint for minimally invasive robotic surgery,” Proc. 2013 IEEE Int. Conf. on Robotics and Automation (ICRA), Karlsruhe, Germany, May 2013, pp. 5807–5812. doi: 10.1109/ICRA.2013.6631364.

J. S. Sandoval Arévalo, G. Poisson, and P. Wieyers, “A new kinematic formulation of the RCM constraint for redundant torque-controlled robots,” Proc. 2017 IEEE/RSJ Int. Conf. on Intelligent Robots and Systems (IROS), Vancouver, Canada, Sept. 2017, pp. 1–6. doi: 10.1109/IROS.2017.8206265.

License

Model assets: see models/LICENSE (original license from the Panda model source).