CLeAR Lab

Collaborative, Learning, and Adaptive Robots Lab at NUS.
We Develop Physical and Social Intelligence for Trustworthy Robots.

GeNIE for In-the-Wild Navigation

We present a robust navigation framework designed for global deployment that achieved first place in the Earth Rover Challenge at ICRA 2025.

Octopi: Object Property Reasoning with Large Tactile-Language Models

We introduce PhysiCLeaR, an annotated dataset of everyday objects and tactile readings collected from a Gelsight Mini sensor, as well as Octopi, a system that leverages both tactile representation learning and large vision-language models to perform physical reasoning and inference, given tactile videos of multiple objects.

Don’t Start from Scratch: Behavioral Refinement via Interpolant-based Policy Diffusion

We develop interpolant policies that leverage informative source distributions for imitation learning.

GRaCE: Balancing Multiple Criteria to Achieve Stable, Collision-Free, and Functional Grasps

We propose an optimization-based grasp synthesis framework, GRaCE, to generate context-specific grasps in complex scenarios. We test GRaCE in a simulator and a real-world grasping tasks.

Tas wins CS Best PhD Thesis Award

Congratulations to Tasbolat Taunyazov for winning the NUS School of Computing Best PhD Thesis Award in CS! The prize is awarded to “the most outstanding PhD ...

ICRA'25 Keynote

Harold gave a keynote at ICRA’25 on a foundation model for agents in embodied AI systems. Check it out below or on youtube: https://youtu.be/NCkwz0dBmO4?si=R...

Harold delivers ICRA'24 Keynote

Harold gave a well-received Keynote at ICRA’24 on our group’s work on generative modeling and robotics! He spoke about two of our recent works on new kinds o...

4 Papers at R:SS'24.

All four of our submissions (3 papers, 1 demo) were accepted to R:SS 2024! A fantastic accomplishment by our CLeAR members Kaiqi, Samson, Tasbolat, Linh, Kel...

Arena 5.0 – A Photorealistic ROS2 Simulation Framework for Social Navigation with NVIDIA Isaac Gym

We introduce Arena 5.0, the fifth iteration of the Arena platform, featuring photorealistic simulation via NVIDIA Isaac Gym, comprehensive benchmarking, and customizable scenario generation!

Arena 4.0 - A Comprehensive ROS2 Development and Benchmarking Platform for Human-centric Navigation Using Generative-Model-based Environment Generation

We introduce the fourth iteration of the Arena platform - Arena 4.0, a platform to develop, train, and benchmark social navigation approaches with generative model–based environment generation!

DISCO: Language-Guided Manipulation with Diffusion Policies and Constrained Inpainting

DISCO uses vision-language models to guide diffusion policies with optimized keyframe inpainting, enabling superior zero-shot open-vocabulary robotic manipulation.

Coupled Hierarchical Diffusion for Long-Horizon Tasks

We propose CHD, a unified diffusion-based framework that tightly couples sub-goal and trajectory generation to improve long-horizon task planning.

VLA-Touch: Enhancing Vision-Language-Action Models with Dual-Level Tactile Feedback

We present VLA-Touch, a framework for VLA models with dual-level tactile feedback for contact-rich manipulation.

Diffusion Meets Options: Hierarchical Generative Skill Composition for Temporally-Extended Tasks

DOPPLER is a new framework that combines diffusion models and hierarchical reinforcement learning to let robots plan and replan complex, long-horizon tasks from offline data with robustness in the real world.