About

Simulation
Real humanoid
TACT-ful

I want humanoids that can take a task, however it's specified, and carry it out in whatever scene they're in. My research is on whole-body control grounded in perception, where the task's intent stays fixed and the execution adapts to what the robot sees.

At VinRobotics I lead the Reinforcement Learning Department. We've taken perceptive locomotion and whole-body compliance from simulation onto real humanoids under load, and I'm now building perception into the pretraining of general whole-body controllers.

I did my MSE in Robotics at Penn and a structured-light vision thesis at HCMUT. I'm applying to PhD programs in robot learning, and I'm open to full-time roles and research collaborations in the meantime.

Outside of work, I travel a lot. More in Travel.

News

Last touched October 2026

  • TACT-ful was accepted to the IROS 2026 Workshop on Full-Shift Robot Co-Workers.
  • Trying to get my hands on a Microduck from Pollen Robotics, say hi if you already have one.
  • Won Second Prize at the Vin33 Chinese Chess Tournament, August 2026. More in Hobbies.

Research

Research map: a task and perception feed a whole-body controller that runs on a real humanoid, with each project placed on the part it addressesTaskhowever it’s specifiedPerceptionterrain, contact, loadWhole-body controllerintent fixed, execution adaptsReal humanoidsim-to-real, under loadcontact forcesTD-MTPMoE terrainTACT-fulCompliantWBCSOFIECurrent work Research map: a task and perception feed a whole-body controller that runs on a real humanoid, with each project placed on the part it addressesTaskhowever it’s specifiedPerceptionterrain, contact, loadWhole-body controllerintent fixed, execution adaptsReal humanoidsim-to-real, under loadcontact forcesTD-MTPMoE terrainTACT-fulCompliantWBCSOFIECurrent work
Current project profile
Three simulated humanoids, each read by a height scan drawn as dots, crossing stepping stones, climbing stairs and walking over rough ground

Pretraining Foundation Policies for Perceptive Humanoid Locomotion with Offline RL Under review

SOFIE learns perceptive humanoid locomotion entirely offline, from rollouts of experts of mixed quality. It scores whole footsteps by their advantage and imitates the better ones more, so one policy walks three humanoids over stepping stones, stairs and rough ground where behavior cloning copies the mistakes. Fine-tuned on a small dataset from an unseen humanoid, a policy pretrained with SOFIE also beats training from scratch.

Project page →
Per-task panels comparing planner alone, multi-step target alone, and both against each task's baseline, with the stair task highlighted where only the combination finishes above baseline

TD-MTP: Coupling Structured Planning with Multi-Step Value Learning Under review

Model-based RL agents usually tune the planner and the critic's learning target separately. Across ten tasks, TD-MTP finds that the two interact. On a humanoid stair task, each change alone lowers the return, and the two combined beat the baseline.

Project page →
Four side-by-side simulation views of the humanoid crossing a gap and clearing a hurdle, with foothold planner targets overlaid on each landing site

Terrain Traversal Without Motion Priors: Expert-Routed Humanoid Locomotion at Deployment Fidelity Under review

Perceptive humanoid locomotion without the usual motion-capture and distillation pipeline. A terrain classifier routes a five-expert mixture over a sensed elevation grid, and every foothold target is refined by closed-form gradient steps on an interpolated terrain-cost field rather than snapped to a fixed grid. Crosses an 85 m five-segment waypoint course 18/20 end-to-end in a second, deployment-fidelity simulator.

Project page →
Service humanoid traversing structured terrain while carrying a payload, guided by terrain affordance cost maps

TACT-ful: Multi-Channel Terrain Affordance and Compliance Training for Payload-Robust Perceptive Humanoid Locomotion IROS 2026 Workshop

A perceptive locomotion system for payload-robust humanoid walking. Multi-channel terrain cost maps guide a GPU-parallel foothold planner, with virtual wrench injection enabling payload-compliance training without force sensors. Achieves 1.0 m/s on stairs with 20 cm risers and zero-shot transfer to a physical humanoid carrying up to 20 kg.

Project page →
Heavy humanoid yielding compliantly to external forces at arbitrary contact sites across wiping, squatting, and cooperative transport tasks

Whole-Body Compliance for Heavy Humanoids via Force Latent Estimation and Residual Impedance Targets Under review

Whole-body compliance for a 70 kg heavy humanoid under high-payload and external-force scenarios. A multi-site impedance reference controller, variational force latent encoder, and bounded residual mechanism jointly edit impedance equilibria, validated on surface wiping, force reaction, loaded squatting, and cooperative transport with a 0.98 success rate.

Project page →

Publications

Under Review

  1. Tai Hoang, Huy Le, Thanh Ly, Truong-Duy Dang, Chien Le, Aleksandar Taranovic, Gerhard Neumann, An T. Le, and Vien Anh Ngo, “Pretraining Foundation Policies for Perceptive Humanoid Locomotion with Offline RL.” Manuscript under review. Project page
  2. Quynh Anh Huynh, Thanh Ly, Hoang M. Truong, Cuc T. Trinh, Chien Le, Georgia Chalvatzaki, and An T. Le, “TD-MTP: Coupling Structured Planning with Multi-Step Value Learning.” Manuscript under review. Project page
  3. Thanh Ly, Chien Le, Truong-Duy Dang, Duy Nguyen Ho Minh, Duy-Dung Le, Vien Anh Ngo, and An T. Le, “Terrain Traversal Without Motion Priors: Expert-Routed Humanoid Locomotion at Deployment Fidelity.” Manuscript under review. Project page
  4. Tan-Dzung Do, Cuc T. Trinh, Phuong Tuan Dat, Truong-Duy Dang, Chien Le, Thanh Ly, Vien Anh Ngo, and An T. Le, “Whole-Body Compliance for Heavy Humanoids via Force Latent Estimation and Residual Impedance Targets.” Manuscript under review. Project page

Workshop Papers

  1. Thanh Ly, Truong-Duy Dang, Chien Le, Tan-Dzung Do, Phuong Tuan Dat, Cuc T. Trinh, Vien Anh Ngo, and An T. Le, “TACT-ful: Multi-Channel Terrain Affordance and Compliance Training for Payload-Robust Perceptive Humanoid Locomotion,” IROS 2026 Workshop on Full-Shift Robot Co-Workers, 2026. Project page

Peer-Reviewed

  1. P. Thanh Ly, Q. Chi Nguyen, N. Duy Hung Nguyen, P.-T. Pham, and K.-S. Hong, “Structured-Light-Based 3D Scanning System for Industrial Manipulator in Bin-Picking Application,” Australian & New Zealand Control Conference (ANZCC), 2022, pp. 34–39.
  2. N. Duy Hung Nguyen, P.-T. Pham, P. Thanh Ly, L. H. Nguyen, and Q. Chi Nguyen, “Bin-Picking Solution for Industrial Robots Integrating a 2D Vision System,” International Conference on High Performance Big Data and Intelligent Systems (HDIS), 2022, pp. 266–270.

Videos

Perceptive Locomotion for Payload-Robust Walking

Humanoid Platform: Global Debut

Load-Bearing Locomotion: Heavy Payload Transport

Dexterous Manipulation: Human–Robot Collaborative Painting

Stereo Vision Bin-Picking (Undergraduate Thesis)