Robotic Systems Lab at TU Wien Uses MATLAB for Real-Time Testing
Slippage-Free Impacts and Hardware-in-the-Loop Experiments Are Realized with Franka Robotics
“The Franka Toolbox for MATLAB became very useful because it allows us to transfer algorithms we developed in theory and in simulations directly on the real robot.”
Key Outcomes
- MATLAB and Simulink tools enabled reliable and effective communication with physical robots
- Researchers created a seamless transition between robotic simulations and control of a physical robot using MATLAB
- Simulation and communication with physical robots using MATLAB resulted in time savings, allowing researchers to focus on developing control systems
Using MATLAB and Simulink tools, researchers conducted hardware-in-the-loop simulations of a suspended multirotor manipulation platform.
Franka Robotics is a German company that focuses on developing robotic platforms and systems, including an articulating robotic arm. To help research customers transition between computer simulations and physical, real-time testing, the Franka team developed Franka Toolbox for MATLAB®. Led by Professor Christian Ott, researchers at the Robotic Systems Lab at TU Wien in Austria apply these tools to complex scenarios in their labs.
In one scenario, a Franka FR3 robot is utilized for hardware-in-the-loop simulation testing of a complex robotic system, consisting of a manipulator attached to a suspended multirotor platform. Researchers used MATLAB and Simulink® to fully simulate the system’s dynamics, with the Franka robot emulating the manipulator’s end-effector dynamics to provide real contact force feedback. The Franka Toolbox for MATLAB facilitated this process, allowing real-time interaction and monitoring of the entire system via Simulink.
The team also applied the toolbox to study post-impact slippage of the Franka robot’s end effector. Suppressing such contact slippage is essential for robotic applications such as hammering or stamping. TU Wien researchers showed that slippage can be prevented if the approach direction aligns with the target’s unique non-slippage impact direction (NSID), which varies due to the robot’s inertial and kinematic properties. Optimizations can identify suitable target points—for example, allowing for vertical, slip-free impacts.
Model-based analyses for this project were conducted using MATLAB and Simulink, with results validated with the Franka Toolbox for MATLAB. Researchers found the seamless transfer from simulation to physical testing and the ability to easily implement user interfaces in Simulink beneficial. The team created an interactive demo to illustrate the variation of the NSID with the target. Users can guide the robot to a desired target while observing the corresponding NSID in a live animation of the robot. Once the target is set, impacts can be performed, demonstrating that slippage is only suppressed under the NSID.
The Franka Toolbox for MATLAB lets researchers control Franka robots directly from Simulink, thus benefitting from MATLAB and Simulink features like efficient control system implementation, online parameter tuning, and quick user interface development. It facilitates seamless transitions between robotic simulations in Simulink and real-world testing, enhancing collaboration between teams and giving them more time to focus on developing control systems for their applications.
In the future, the team is interested in exploring how MATLAB and Simulink can be used for further robotic hardware.
