Scholar
Tao Pang
Google Scholar ID: BNNNS-wAAAAJ
Roboticist, RAI Institute
Citations & Impact
Citations
456
H-index
10
i10-index
11
Publications
20
Co-authors
26
list available
Publications
Resume (English only)
Academic Achievements
- “Dexterous Contact-Rich Manipulation via the Contact Trust Region” (submitted, 2025)
- “Global Planning for Contact-Rich Manipulation via Local Smoothing of Quasi-dynamic Contact Models,” IEEE Transactions on Robotics (T-RO), 2023 — Honorable Mention for IEEE T-RO King-Sun Fu Memorial Best Paper Award
- “Bundled gradients through contact via randomized smoothing,” IEEE Robotics and Automation Letters (RA-L), 2022
- “Should We Learn Contact-Rich Manipulation Policies from Sampling-Based Planners?,” IEEE RA-L, 2025
- “Physics-Driven Data Generation for Contact-Rich Manipulation via Trajectory Optimization,” Robotics: Science and Systems (RSS), 2025
- “Planning-Guided Diffusion Policy Learning for Generalizable Contact-Rich Bimanual Manipulation” (submitted, 2024)
- “A convex quasistatic time-stepping scheme for rigid multibody systems with contact and friction,” ICRA 2021
- “A robust time-stepping scheme for quasistatic rigid multibody systems,” IROS 2018
- Collaborated with R. Tedrake, H.J.T. Suh, T. Zhao, and others on multiple publications
Research Experience
- Roboticist at the Robotics and AI Institute (formerly Boston Dynamics AI Institute)
- Conducts research on global planning for contact-rich manipulation using smoothing, quasi-dynamic contact models, and classical motion planning
- Investigates quasi-static rigid body dynamics for efficient simulation and planning in contact-rich tasks
- Explores whole-body contact force estimation and control using only joint torque measurements to enhance safety during accidental collisions
- Develops planning-guided behavior cloning and diffusion policy learning for generalizable bimanual manipulation
Background
- Roboticist at the Robotics and AI Institute (formerly Boston Dynamics AI Institute)
- Research aims to enable robots to confidently and intelligently make contacts
- Interested in enabling robots to intelligently and dexterously manipulate objects and surroundings with rich contact mirroring human cadence
- Focuses on: (i) efficient global planning for contact-rich manipulation by leveraging contact model structure; (ii) imitation learning from planner-generated data
- Believes model-based reasoning efficiency is crucial for generating large-scale datasets needed for robotics foundation models with robust, generalizable, and dexterous interaction capabilities
