TVL (Touch-Vision-Language) dataset

Name: TVL (Touch-Vision-Language) dataset
Creator: UC Berkeley, Meta AI, TU Dresden, CeTI
Keywords: research-lab

TVL (Touch-Vision-Language) dataset is a language and reasoning annotation robotics dataset maintained by UC Berkeley, Meta AI, TU Dresden, CeTI. Primary hardware platform: handheld 3D printed device with DIGIT sensor and Logitech BRIO webcam. Access is open under not stated.

Dataset specifications
Year	2024
Embodiments	handheld 3D printed device with DIGIT sensor and Logitech BRIO webcam
Task categories	manipulation
License	not stated
Access	open
Maintainer	UC Berkeley, Meta AI, TU Dresden, CeTI