Harold Soh
Associate Professor
Directs the CLeAR group on machine learning and decision-making for trustworthy collaborative robots, from human trust modelling to tactile e-skins.
Harold Soh's lab at NUS works on human-robot interaction & assistive robots, human-AI collaboration & trust and reinforcement learning. 31 papers since 2023, 26 at top venues, mostly RSS and ICRA. 10 PhD students. 24 alumni, most often to Light Robotics and Amazon. Applications close .
At a glance
- papers, 2023–2026
- 8/ year
- at T1 venues
- 84%
- PhD students
- 10
- MS students
- 0
- alumni listed
- 24
Papers by yearfrom the lab homepage, 2023–2026
- 7 papers2026 · T1 6 · collecting
- 8 papers2025 · T1 6
- 9 papers2024 · T1 7
- 7 papers2023 · T1 7
T1 venuesOtherVenues: RSS, ICRA, NeurIPS, RA-L, IROS
Research focuspaper titles that use each topic's words
- Human-robot interaction & assistive robots4 papers
- Human-AI collaboration & trust3 papers
- Reinforcement learning3 papers
- tactile2 papers
- trust0 papers
Paper list
From the lab homepage.
20267 papers
- ICLR (T1)Abstracting Robot Manipulation Skills via Mixture-of-Experts Diffusion Policies (opens in a new tab)
- ICML (T1)Conflict-Aware Additive Guidance for Flow Models under Compositional Rewards (opens in a new tab)
- RSS (T1)Guided Streaming Stochastic Interpolant Policy (opens in a new tab)
- NeurIPS (T1)Not All Layers Need Tuning: Selective Layer Restoration Recovers Diversity (opens in a new tab)
- RSS (T1)SkillVLA: Tackling Combinatorial Diversity in Dual-Arm Manipulation via Skill Reuse (opens in a new tab)
- ICRA (T1)TOPO-Bench: An Open-Source Topological Mapping Evaluation Framework with Quantifiable Perceptual Aliasing (opens in a new tab)
- RA-LVLA-Touch: Enhancing Vision-Language-Action Models with Dual-Level Tactile Feedback (opens in a new tab)
20258 papers
- ICRA (T1)Arena 4.0: A Comprehensive ROS2 Development and Benchmarking Platform for Human-centric Navigation Using Generative-Model-based Environment Generation (opens in a new tab)
- RSS (T1)Arena 5.0: A Photorealistic ROS2 Simulation Framework for Developing and Benchmarking Social Navigation (opens in a new tab)
- CoRL (T1)CHD: Coupled Hierarchical Diffusion for Long-Horizon Tasks (opens in a new tab)
- ICRA (T1)Diffusion Meets Options: Hierarchical Generative Skill Composition for Temporally-Extended Tasks (opens in a new tab)
- RA-LDISCO: Language-Guided Manipulation with Diffusion Policies and Constrained Inpainting (opens in a new tab)
- RA-LGeNIE: A Generalizable Navigation System for In-the-Wild Environments (opens in a new tab)
- ICRA (T1)Imitation Learning with Limited Actions via Diffusion Planners and Deep Koopman Controllers (opens in a new tab)
- ICLR (T1)Know When to Abstain: Optimal Selective Classification with Likelihood Ratios (opens in a new tab)
20249 papers
- RSS (T1)Arena 3.0: Advancing Social Navigation in Collaborative and Highly Dynamic Environments (opens in a new tab)
- RSS (T1)Don’t Start from Scratch: Behavioral Refinement via Interpolant-based Policy Diffusion (opens in a new tab)
- Joint European Conference on Machine Learning and Knowledge Discovery in DatabasesGenerative Modeling with Flow-Guided Density Ratio Learning (opens in a new tab)
- RSS (T1)GRaCE: Balancing Multiple Criteria to Achieve Stable, Collision-Free, and Functional Grasps (opens in a new tab)
- RA-LLTLDoG: Satisfying Temporally-Extended Symbolic Constraints for Safe Diffusion-based Planning (opens in a new tab)
- RSS (T1)Octopi: Object Property Reasoning with Large Tactile-Language Models (opens in a new tab)
- NeurIPS (T1)Out-of-Distribution Detection with a Single Unconditional Diffusion Model (opens in a new tab)
- ICRA (T1)Probable Object Location (POLo) Score Estimation for Efficient Object Goal Navigation (opens in a new tab)
- TMLR (T1)Stochastic Bandits for Egalitarian Assignment (opens in a new tab)
20237 papers
- IROS (T1)Large Language Models as Zero-Shot Human Models for Human-Robot Interaction (opens in a new tab)
- IROS (T1)Latent Emission-Augmented Perspective-Taking (LEAPT) for Human-Robot Interaction (opens in a new tab)
- ICML (T1)Neural Continuous-Discrete State Space Models for Irregularly-Sampled Time Series (opens in a new tab)
- IROS (T1)Refining 6-DoF Grasps with Context-Specific Classifiers (opens in a new tab)
- ICRA (T1)Safety-Constrained Policy Transfer with Successor Features (opens in a new tab)
- NeurIPS (T1)Selective Amnesia: A Continual Learning Approach to Forgetting in Deep Generative Models (opens in a new tab)
- NeurIPS (T1)The Best of Both Worlds in Network Population Games: Reaching Consensus and Convergence to Equilibrium (opens in a new tab)
Where alumni went24 listed, 59% to industry
Industry 10Academia 7
Most often to
- Light Robotics
- Amazon
- Espressif Systems
Also to
- NUS
- Carnegie Mellon University
- Zhongguancun Academy
Apply to NUSFall 2027
School of Computing PhD in Computer Science
- Round
- Main
- Opens
- —
- GRE
- See program page
- English
- See program page
- Fee
- See program page
Not yet confirmed for this intake.
Similar labsnearest by research summary
- GBOxford
Jack Umenberger
- Reinforcement learning
- USGeorgia Tech
Matthew Gombolay
- Human-robot interaction & assistive robots
- USUIUC
Katherine Driggs-Campbell
- Manipulation & robot learning
- GBImperial
Yiannis Demiris
- Human-robot interaction & assistive robots
- USMichigan
Lionel Robert
- Human-robot interaction & assistive robots
- CNSJTU
Lipeng Chen
- Manipulation & robot learning
Your notes
Save this lab to keep notes.
Data
Checked against the lab homepage.
- Collected
- Verified
- Papers
- The lab homepage
- Members
- Lab homepage
- Alumni
- Lab homepage
Tiers are venue classifications, not lab rankings.
NUScloses