Shaoxiong (Shawn) Wang

Roboticist @ Dexterity

Email: shawn.wang [at] dexterity (dot) ai
Google Scholar / CV

I work on Contact-Rich Manipulation at Dexterity, a robotics startup based in the Bay area, focusing on intelligent robots for Logistics.


Before joining Dexterity, I finished my Ph.D. in the Computer Science and Artificial Intelligence Laboratory (CSAIL) at Massachusetts Institute of Technology (MIT) supervised by Prof. Edward H. Adelson. I also had the opportunity to work with Prof. Alberto Rodriguez. I recieved my B.E. degree from Department of Computer Science and Technology, Tsinghua University.


During Ph.D., I studied Tactile Perception & Manipulation, Computer Vision, and Machine Learning.


Previously, I was fortunate to work in Knowledge Engineering Group, Tsinghua University, advised by Prof. Jie Tang; Facebook AI Research, advised by Dr. Roberto Calandra.




News


Publications


Visuotactile Affordances for Cloth Manipulation with Local Control

CoRL’22
[Paper] [Project Page]


See, Hear, Feel: Smart Sensory Fusion for Robotic Manipulation

CoRL’22
[Paper] [Project Page]


Towards Learning to Play Piano with Dexterous Hands and Touch

IROS’22
[Paper]


TACTO: A Fast, Flexible, and Open-Source Simulator for High-Resolution Vision-Based Tactile Sensors

RA-L and ICRA’22
(NeurIPS'20 Workshop, Robot Learning; Deep Reinforcement Learning)
[Paper] [Code]


GelSight Wedge: Measuring High-Resolution 3D Contact Geometry with a Compact Robot Finger

ICRA'21
[Paper] [Project Page]


PyTouch: A Machine Learning Library for Touch Processing

ICRA'21
[Paper] [Code]


SwingBot: Learning Physical Features from In-Hand Tactile Exploration for Dynamic Swing-up Manipulation

IROS'20 Best Paper Award
[Paper] [Project Page] [MIT CSAIL News]


Cable Manipulation with a Tactile-Reactive Gripper

RSS'20 Best Paper Award Finalist
IJRR'21
[Paper] [Project Page] [MIT News]


3D Shape Perception from Monocular Vision, Touch, and Shape Priors

IROS'18
[Paper] [Project Page]


Active Clothing Material Perception using Tactile Sensing and Machine Learning

ICRA’18
[Paper][Data] [Project Page]



Connecting Look and Feel: Associating the visual and tactile properties of physical materials

CVPR’17 Oral
[Paper][Data][Project Page]


Active Zero-Shot Learning

CIKM’16
[Paper][Data]

Side Projects


MIT 6.943[J] How To Make (Almost) Anything

Teleoperation with force feedback.

Features:

         Micro-controller : ATmega16U4

         USB-support : Based on LUFA USB library

         Force Sensing : Hall-effect sensors & magnet

         3D-Printed : Cable-driven robot hand, Yale Openhand
[Video] [Website]


Music Generation with Recurrent Neural Network

1st Place in Microsoft Campus Elite Program  News
Grand Prize in 35th Tsinghua Challenge Cup (top 6 among 381 teams)   News

A comprehensive music generation system based on recurrent neural network. It generates music automatically driven by data and is embedded with rich music knowledge.



Shaoxiong (Shawn) Wang 2022