Few Remarks on Benchmarks
Presented at the Workshop in Benchmarking Robotics
Bayesian Optimization for Robotics
Designing and tuning controllers for real-world robots is a daunting task which typically requires significant expertise and lengthy experimentation. Bayesian optimization has shown to be a successful approach to automate these tasks with little human expertise required. In this talk, I will discuss the main challenges of robot learning, and how BO helps to overcome some of them. Using as showcase real-world applications where BO proved to be effective, I will also discuss how the challenges encountered in robotics applications can guide the development of new BO algorithms.
Learning Model-based Control for (Aerial) Manipulation
In this talk, I will discuss what I consider two of the crucial challenges of manipulation: the use of tactile sensors, and model-based control. Humans make extensive use of touch, but integrating the sense of touch in robot control has traditionally proved to be a difficult task. As an alternative, we propose the use of data-driven machine learning methods, to learn complex multi-modal models from raw sensor measurements. I will conclude by discussing useful lessons we learned along the way, and how some of these lessons could be valuable in the context of aerial manipulation.
Robots and the Sense of Touch
Humans make extensive use of touch. However, integrating the sense of touch in robot control has traditionally proved to be a difficult task. In this talk, I will discuss how machine learning can help to provide robots with the sense of touch, and the benefits of doing so.