Safety and optimality in learning-based control at low computational cost
It addresses safety and computational efficiency for real-world robotic applications, offering a practical solution but appears incremental as it builds on existing safe learning methods.
The paper tackles the problem of providing safety guarantees in learning-based control for physical systems without high computational cost, proposing CoLSafe, a lightweight algorithm with sublinear complexity that demonstrates effectiveness on a seven-degrees-of-freedom robot arm.
Applying machine learning methods to physical systems that are supposed to act in the real world requires providing safety guarantees. However, methods that include such guarantees often come at a high computational cost, making them inapplicable to large datasets and embedded devices with low computational power. In this paper, we propose CoLSafe, a computationally lightweight safe learning algorithm whose computational complexity grows sublinearly with the number of data points. We derive both safety and optimality guarantees and showcase the effectiveness of our algorithm on a seven-degrees-of-freedom robot arm.