Reproduce “Velocity-Form Data-Enabled Predictive Control of Soft Robots under Unknown External Payloads” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Huanqing Wang, Kaixiang Zhang, Kyungjoon Lee, Yu Mei, Vaibhav Srivastava, Jun Sheng, Ziyou Song, Zhaojian Li
Researchers from Michigan State University, University of California, Riverside, and the University of Michigan developed a velocity-form Data-Enabled Predictive Control (VDeePC) framework for soft robots, allowing robust operation under unknown external payloads using only unloaded training data. The VDeePC approach reduced steady-state tracking error by 89% compared to standard DeePC and maintained Root Mean Square Error below 0.42 mm across various unmodeled payloads.
Key insights
- Adapting the Data-Enabled Predictive Control (DeePC) framework to a 'velocity-form' intrinsically incorporates offset-free control properties, providing robustness against unknown constant or slowly varying disturbances like external payloads.
- The method leverages Willems’ Fundamental Lemma to achieve model-free control by directly using input-output trajectories, eliminating the need for explicit system identification or basis function selection.
- Combining data-driven predictive control with incremental dynamics enables accurate soft robot control under unknown external payloads using only data collected in unloaded conditions, significantly improving data efficiency.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.