Towards the best robot for the job: Optimising actuation design through multi-task co-design and component selection
Wesley Roozing, Jonathan Cornee Schaaij, Alessandro Forino
AI summary
Problem
Most robot co-design methods focus on single tasks or struggle to translate continuous optimization results into practical, discrete component selections from manufacturer catalogs.
Approach
The authors formulate a trajectory optimization framework that simultaneously tunes motor sizes and gear ratios for a representative set of tasks using a data-driven actuation model, then maps the results to discrete catalog components.
Key results
- Multi-task co-design reduces average energy use by 23% and task duration by 14% over trajectory optimization alone
- A unified design matches the average performance of robots individually optimized per task
- A practical pipeline successfully selects feasible discrete motor-gear combinations from catalogs
- Hyperparameter analysis quantifies the impact of task weighting and collocation density on design outcomes
Why it matters
Enables roboticists to rapidly design highly efficient, multi-purpose manipulators using off-the-shelf components without iterative trial-and-error.
Abstract
We propose a multi-task co-design approach to design a robot’s actuation (motor sizes and gear ratios) based on trajectory optimisation. Leveraging an actuation model fit on data of series of components, we find the optimal set of design parameters for all joints over a set of representative tasks for the given robot. Critically, we close the loop towards component selection, given a finite set of available components. This enables more practical use of co-design tools. Our results show that the method is effective, and critically, show that it is possible to find a robot design that is capable of performing an entire set of tasks at an efficiency that is comparable to a robot co- designed for each specific task. Finally, we perform an extensive analysis of hyperparameter effects, and select discrete actuation components from catalogues and compare to co-design results.