About
Recent developments in design and control of torque controlled actuators have already lead to remarkable advancements in safety, robustness and interaction performance for torque controlled robots and assistive robotic devices. Still, the actuator development for the above robots and robotic devices relies heavily on the intuition and experience of an engineer than on any rigorous theory. In the author's opinion, literature lacks proper understanding of relevant requirements along with metrics for their quantification to guide this process. Conventional notions (power-density, peak torque, maximum speed, single numbers for bandwidths etc.) are insufficient for new applications that are dominated by physical interaction.
The Compliant Joint Toolbox is available under GPL-3.0. It has emerged during the author’s work on actuator modelling, design and control towards solutions for the above actuator design challenges in diverse robotic applications. The toolbox has helped the authors to cope with the variety of actuator configurations they had to consider in terms of e.g. rotor, gearbox and torque sensor combinations during the development of the WALK-MAN, CENTAURO and CogIMon robots.
| WALK-MAN |  |
|---|---|
| CENTAURO |  |
The Compliant Joint Toolbox supported the study of the dynamics and control of different compliant electrical actuators with integrated torque sensors across arbitrary parameter ranges and investigate the impact of nonlinear phenomena such as friction and ripple. This way, the toolbox has been an essential tool in the preparation of publications modelling, observer design, torque and impedance control. Through this process, the code has reached a certain level of maturity that permits productive use.
The authors hope Compliant Joint Toolbox can catalyse the ongoing discussion on compliant robot actuation, support academic education in the field and contribute to the propulsion of community efforts towards common notions, metrics and benchmarks that ease torque controlled actuation design, comparison and selection across diverse robotic applications. Hence, the motivation to make Compliant Joint Toolbox public.
The Compliant Joint Toolbox is implemented in MATLAB/Simulink, which is a proprietary software suite for technical computing and rapid algorithm prototyping. The MATLAB language is interpreted, requires low learning efforts, ships with numerous state-of-the-art algorithms and visualization tools and therefore has a short time to productivity.
The Python language shares most of the features above. Being non-proprietary, Python would have been the author’s preferred choice to implement Compliant Joint Toolbox and make it available to the community entirely for free. However, the crucial aspect that has triggered the decision against a purely non-proprietary solution is the lack of a mature and sufficiently powerful open alternative for the features offered by the Simulink Real-Time Toolbox. It offers the chance to quickly interface with the actuator hardware based on standard industrial protocols. This allows to rapidly develop, deploy, tune and test models and controllers on different actuator hardware and even with the actuator hardware-in-the-loop. Minimizing the time and effort required to port developed concepts from simulation to experiments, improves realism in research.
The authors, Jörn Malzahn and Wesley Roozing, are Post-Docs at the Italian Institute of Technology in Genoa, Italy.
If you are using the Compliant Joint Toolbox in your work, please reference it accordingly:
| Citation ( BibTeX ) | |
|---|---|
 |
Malzahn J., Roozing W., Tsagarakis N.: “The Compliant Joint Toolbox for MATLAB: An Introduction with Examples. IEEE, submitted, 2018. |
The following works have used and contributed to the current state of the Compliant Joint Toolbox:
-
Baccelliere, Lorenzo, et al. "Development of a human size and strength compliant bi-manual platform for realistic heavy manipulation tasks." Intelligent Robots and Systems (IROS), 2017 IEEE/RSJ International Conference on. IEEE, 2017. [link]
-
Malzahn, Jörn, et al. "What is the torque bandwidth of this actuator?." Intelligent Robots and Systems (IROS), 2017 IEEE/RSJ International Conference on. IEEE, 2017. [link]
-
Roozing, Wesley, et al. "On the Stiffness Selection for Torque-Controlled Series-Elastic Actuators." IEEE Robot. Autom. Lett 2 (2017): 2255-2262. [link]
-
Tsagarakis, Nikos G., et al. "WALK‐MAN: A High‐Performance Humanoid Platform for Realistic Environments." Journal of Field Robotics 34.7 (2017): 1225-1259. [link]
-
Roozing, Wesley, et al. "Comparison of open-loop and closed-loop disturbance observers for series elastic actuators." Intelligent Robots and Systems (IROS), 2016 IEEE/RSJ International Conference on. IEEE, 2016. [link]
This work has received financial support by the European Research Council projects WALK-MAN (no. 611832), CENTAURO (no. 644839) and CogIMon (no. 644727).
