Machine learning and tactile sensing applied to dexterous manipulation
Dexterous manipulation is a crucial ability for the next generation of robots. On this project, a robotic hand equipped with multi-modal tactile sensors provided perception information used to train machine learning models1. This setup was able to provide insights into robotic in-hand object characterization.
This project also involves teleoperated human-robot interaction for in-hand manipulation, such as demonstrated in the following video.
Prado da Fonseca, V., Alves de Oliveira, T. E., & Petriu, E. M. (2019). Estimating the Orientation of Objects from Tactile Sensing Data Using Machine Learning Methods and Visual Frames of Reference. Sensors, 19(10), 2285. https://doi.org/10.3390/s19102285 ↩