Improving Robotic Assistant Dexterity

Project Motivation

As robotic assistants operate in complex, unstructured human-centered environments it becomes essential that robots 

1. Have access to as much useful data as possible as they interact with the environment
2. Be able to leverage that data using intelligent models of the environment 

Challenges for long term deployable robotic systems are 

• Being able to handle uncertainty in a known environment
• Being able to reason about the properties of a new object/element in their environment 
• Recovering from failure to complete a task/objective

By equipping robots to handle these challenges, we can develop robotic platforms that are capable of sustained, reliable, long-term service deployment.

Research Objectives

Research objectives include: 

• Design: Can we leverage anthropomorphic inspiration in a curved fingertip capable of deforming and sensing friction?
• Modeling: Can we efficiently model a continuous curved fingertip with high-resolution (in a computationally tractable way) and characterize stable grasps?
• Motion Planning: Can we leverage these efficient, continuous models of our fingertips to plan multi-finger grasps for in-hand manipulation?

Current Students

• Won Kyung Do
• Bianca Jurewicz
• Camille Chungyoun
• Jose Solano
• Karina Ting
• Tejas Yogesh
• Ankush Dhawan
• Mathilda Kitzmann

Related Publications

• DenseTact 2.0: Optical Tactile Sensor for Shape and Force Reconstruction (preprint)
• DenseTact: Optical Tactile Sensor for Dense Shape Reconstruction