Panda-Base Digital Twin — Modular Simulation of a High-Moment Manipulation Setup

1. Overview & Objective

Panda-Base Digital Twin is a Gazebo/RViz simulation that mirrors a Franka Emika Panda arm mounted on a custom aluminium-profile base used at TUM-MIRMI. The goal is to validate high-moment manipulation controllers and sensor placements entirely in software—eliminating risky, time-consuming on-robot trials while achieving real-time performance.

2. Features & Design Highlights

Full-physics Gazebo 11 world of base + Panda arm + FTE-Omega-160-IP60 SI-1000-120 6-DoF force-torque sensor stub with accurate inertias and collision geometry.
Modular Xacro/URDF stack—common, base, arm, sensor—for rapid reuse or hardware redesign.
Pre-tuned controller suite (joint-position, effort, trajectory, Cartesian-impedance) with gains matched to the twin’s dynamics.
Interactive RViz tele-operation via interactive_marker.py; streams a 6-DOF equilibrium-pose marker at 200 Hz and clamps motion inside safe limits.
One-command startup: roslaunch panda_base_sim panda_base_sim.launch launches Gazebo, controllers, RViz, and the marker node.
Comprehensive documentation & MIT license make it an open template for Panda manipulation research.

3. Architecture & Pipeline

Package structure

panda_base_sim/
├── config/      # controller & hardware YAMLs
├── launch/      # Gazebo / RViz launch files
├── robots/      # Xacro macros: common, base, arm, sensor
├── meshes/      # STL / DAE visuals from SolidWorks
├── scripts/     # interactive_marker.py
└── urdf/        # generated URDFs & CAD-export variants

Launch flow

panda_base_sim.launch
- Loads Gazebo (headless or GUI).
- Generates robot_description from urdf/base_panda.urdf.xacro.
- Pushes controller parameters from config/*.yaml.
- Spawns the robot model into Gazebo at preset joint angles.
- Starts controller_manager spawners (state, trajectory, Cartesian-impedance).
- Optionally launches RViz with robots/panda_base_sim.rviz.
- Runs interactive marker node for live equilibrium-pose commands.
onlybase_gazebo.launch
- Spawns the static base + arm without controllers/RViz for quick visualization.

The force-torque sensor placeholder matches the real FTE-Omega-160-IP60 SI-1000-120 unit used in hardware tests, ensuring mass and inertia fidelity.

Interactive marker pipeline

Gazebo ←→ controller_manager ←→ cartesian_impedance_example_controller
                                       ↑
              equilibrium_pose  (geometry_msgs/PoseStamped)
                                       ↑
                          interactive_marker.py

The Python node reads the current Panda end-effector pose, enforces XYZ bounds, and publishes updates at 200 Hz, enabling smooth, safe tele-op inside RViz.

4. Technologies Used

ROS Noetic (gazebo_ros, controller_manager, franka_ros, interactive_markers)
Gazebo 11 physics engine
Franka Gazebo plugin (FrankaHWSim) for realistic dynamics
RViz with saved display config for live visualization
SolidWorks for CAD design and inertia extraction of the aluminium base
Xacro ⁄ URDF for parameterised robot descriptions
Python 3 (+ rospy) for the interactive marker node
CMake / catkin build system for ROS packaging

5. Workflow & Usage

Clone & build

cd ~/catkin_ws/src
git clone https://github.com/yunusdanabas/panda_base_sim.git
cd ~/catkin_ws
rosdep install --from-paths src --ignore-src -r -y
catkin_make           # or: colcon build
source devel/setup.bash

Run the full twin
```
roslaunch panda_base_sim panda_base_sim.launch
```
This single command starts Gazebo + controllers + RViz and spawns a 6-DOF interactive marker for Cartesian-impedance control.
Minimal spawn
```
roslaunch panda_base_sim onlybase_gazebo.launch
```
Spawns the static base-arm pair in Gazebo without controllers or RViz—ideal for quick visual checks.
Tele-operate
- Drag the Equilibrium Pose marker in RViz for real-time impedance control (clamped workspace).
- Or publish your own poses:
```
rostopic pub /equilibrium_pose geometry_msgs/PoseStamped ...
```

6. Results & Impact

Zero on-robot risk – 100 % of controller tuning and load-case checks completed in simulation.
Real-time performance – ≈ 1 × wall-clock speed (≥ 60 FPS) on a mid-tier desktop with the full controller stack active.
Adopted by peers – 3 MIRMI researchers now use this twin as a baseline for force-torque sensor scaling studies.
Open resource – MIT-licensed code released on GitHub for the robotics community.