UMI-FT

UMI-FT brings together a couple fo projects from Amazon Robotics and Stanford HAI. Both are in collaboration with Prof. Shuran Song (EE) and involve adaptations of her lab's UMI design.

• "In the wild robot data collection with UMI-F/T" is a project with Amazon Lab 126 aimed at developing a new, improved version of UMI that features 6-axis, dynamic force/torque sensing at each fingertip so that finger/object contact forces and internal grasp forces are recorded---along with video, motions from an IMU, and sounds from a contact microphone.
  • Our Amazon Liaison is BDML alumna Dr. Alice Wu.
  • Considerable work is going into upgrading the gripper design and the communications & synchronization capabilities to support this multi-modal data stream.
  • We are also exploring either making a modified open-source version of UMI-F/T that would not be too difficult for others to reproduce, or to manufacturing a limited batch of kits for researchers. Stay tuned...
• In-the-Wild Multimodal Data Collection for Compliant Assistive Robot Skill Acquisition This project will be aimed at using UMI-FT specifically to train tasks relevant for robots in homes of the elderly, helping them to remain independent.