I’m a Robotics master student from Yale University, major in mechanical engineering & material science. My research interest is physics-informed machine learning in robotics, especially differentiable simulator, for contact rich manipulation.

I have divided my researches into four key areas:

• (A) Perception - Infer uncertain environment.

• (B) Reasoning & Task planning – High-level planner.

• (C) Control - Control for contact rich operation.

• (D) Mechanics Design - Co-Design Mechanism with Sensing, Perception, and Control.

Feel free to see my research statement for more details, including future research plans.

Contact: Tianyi.Xiang@yale.edu

My Resume

You can find my CV here:Tianyi’s CV

Publications

"Real-to-Sim via End-to-End Differentiable Simulation and Rendering"

Yifan Zhu, Tianyi Xiang, Aaron Dollar, Zherong Pan

IEEE Robotics and Automation Letters (RA-L 2024), Manuscript submitted for publication.

"A Novel Approach to Grasping Control of Soft Robotic Grippers based on Digital Twin"

Tianyi Xiang, Borui Li, Quan Zhang, March Leach, Enggee Lim

29th International Conference on Automation and Computing (ICAC 2024).

"Development of a Simple and Novel Digital Twin Framework for Industrial Robots in Intelligent Robotics Manufacturing"

Tianyi Xiang, Borui Li, Xiaonan Pan, Quan Zhang

20th International Conference on Automation Science and Engineering (CASE 2024).

"High performance (Zn₀.₅Mg₀.₅)TiO₃ ceramics based composite films for powering multi-mode translation unit and human motion monitoring"

B. Xie, Y. Xie, Y. Ma, N. Luo, Tianyi Xiang, et al.

ACS Applied Materials & Interfaces(2024).

Projects Video & Pictures

A differentiable particle based simulator with gaussian splatting rendering

Inversely constructed of paper: Physically Embodied Gaussian Splatting: A Visually Learnt and Physically Grounded 3D Representation for Robotics

Github Link: https://github.com/Tianyi20/pdb_demo

Real-to-Sim via End-to-End Differentiable Simulation and Rendering

Paper

Liquid Manipulation: Category-level pose & dimension detector with pouring action optimizer

Forked PDDLstream Task and Motion planning (TAMP) online replanning.

(simulation forked from Caelan Reed Garrett LTAMP)

Behaviour cloning (BC) learning-based Block Pushing task

Vision-learning based Soft Actuator Grasping Control within Digital Twin

Simple and Novel Motion Planning Digital Twin Framework for Industrial Robot

Autonomous tracking DIY Mars Rover Design

Tactile Sensing achieving human motion monitoring

ROS SLAM AGV navigation with Adaptive Monte Carlo Localization (AMCL)

AGV with PLC control optimisation simulation