Hi! I'm Pang.

My research aims to enable robots to bravely and intelligently make contacts.

About Me


I’m Tao Pang, a PhD candidate at the Robot Locomotion Group at MIT CSAIL, advised by Prof. Russ Tedrake. I’m interested in designing planning and control algorithms for contact-rich interactions between robots and and their surrounding environment.

Prior to MIT, I received my M.Eng. (EE) from the National University of Singapore (NUS), advised by Prof. Ben M. Chen. I also received my B.Eng. (ME) from NUS, with the Lee Kuan Yew Gold Medal.

CV | Google Scholar | Github | Youtube | LinkedIn


Planning for Contact-rich Manipulation

Combining smoothing, quasi-dynamic models and RRT, our planning-through-contact algorithm can power through local minima and the exponential explosion of contact modes, and generate complex contact-rich manipulation plans such as the one shown on the right.

  • Global Planning for Contact-Rich Manipulation via Local Smoothing of Quasi-dynamic Contact Models, under review. [pdf][video]
    T. Pang*, H.J.T. Suh*, L. Yang, R. Tedrake.
  • Bundled gradients through contact via randomized smoothing, in IEEE Robotics and Automation Letters (RA-L), 2022. [pdf][video]
    H.J.T. Suh*T. Pang*, R. Tedrake.

Quasi-static Rigid Body Dynamics

Rigid body systems are close to force equilibrium for many robotic manipulation tasks (e.g. the one on the right). This assumption allows simulating such systems with much larger time steps, as indicated by the choppiness in the video, than using second-order dynamics. Therefore, quasi-static models are more amenable for planning, as the planner can look further into the future while taking fewer steps.

  • A convex quasistatic time-stepping scheme for rigid multibody systems with contact and friction, in ICRA2021. [pdf][video]
    T. Pang, R. Tedrake.
  • A robust time-stepping scheme for quasistatic rigid multibody systems, in IROS2018. [pdf]
    T. Pang, R. Tedrake.

Contact Force Estimation & Control from Joint Torque

Whole-body contact force estimation using only joint-torque measurements has severe limitations. Nevertheless, under the right circumstances, they can keep both the robot and the environment safe when accidental collision happens. As shown on the right, the robot bumped into an egg, but did not crush it! 

  • Easing reliance on collision-free planning with contact-aware control, in ICRA2022. [pdf][video]
    T. Pang, R. Tedrake.
  • Identifying external contacts from joint torque measurements on serial robotic arms and its limitations, in ICRA2021. [pdf][video]
    T. Pang, J. Umenberger, R. Tedrake.

Gasoline Variable-pitch Quadcopter

A strange creation that weighs 10 kg and can remain airborne for up to 3 hours by burning fossil fuel. It was built at a time when electric quadcopters were getting popular due to their mechanical simplicity, but could only fly for less than 30 minutes. Not any more. 

  • Towards long-endurance flight: Design and implementation of a variable-pitch gasoline-engine quadrotor, in IEEE ICCA 2016. [pdf][video]
    T. Pang, K Peng, F Lin, B.M. Chen.


3D Geometry Reconstruction from NeRF

Class project of MIT 6.869. I observed that in order for NeRF to accurately learn the object geometry, color needs to be a function of both spatial position and viewing direction.

[interactive slides][repo]

the Rendering Pipeline

Homeworks of CMU 15.462. I implemented CPU-based rendering pipelines, including rasterization, mesh manipulation and ray-tracing accelerated by Bounding Volume Hierarchy (BVH). The image to the right is rendered by code written by me! 


Active SLAM for a Planar Robot

Class project of MIT 16.485, where I implemented a path planner that actively seeks to reduce pose uncertainty by returning to observed landmarks, in addition to SLAM with loop-closure. 

[interactive slides][repo]