Research

Dexterous Manipulation Planning

How to make robots as creative as human beings in manipulation? My current research focuses on dexterous manipulation motion generation, the first step towards general manipulation intelligence. We considered the dexterity from planning through both the robot hand contacts and environment contacts. This problem is challenging due to the non-smooth and hybrid nature of contacts.

This is our first step [2D planner] [3D planner]. To the best of our knowledge, our planner is the first method that is capable of solving diverse dexterous manipulation tasks without any pre-designed skill or pre-specified contact modes.

Manipulation with Suction Cups

If there is an octopus civilization, maybe most of its robotic manipulation study will be about suction cups. However, human beings only use vacuum suction cups for pick and place. Suction cups have the potential to be powerful manipulators. We want to explore what other manipulation tasks suction cups can do, and extend our study to soft robotic fingers.

Screwdriving

We are interested in screwdriving for mini-screws. Hundreds of billions of small screws are assembled in consumer electronics industry every year, yet reliable screwdriving automation remains one of the most challenging tasks.

To study the screwdriving processes, we collected a large screwdriving dataset including more than 8,000 screwdriving operations. For more information, please visit our dataset webpage.

We also developed a data efficient failure detection and stage classification framework for screwdriving [here].