Dec-MARVEL: Decentralized Multi-Agent Exploration without Communication under Budget Constraints
Robot teams that coordinate on glimpses of one another rather than messages, while keeping enough travel budget to return to base.
Huawei Singapore · Norbert Wiener Research Centre
Jimmy Chiun
AI Research Scientist
I work at the intersection of embodied AI, reinforcement learning, and robotics. My research focuses on multi-task learning and world models for long-horizon robotic tasks, building agents that reason, adapt, and act reliably in complex environments.
8
papers & preprints
- ICRA
- CoRL
- IEEE RA-L
- IROS
Research Focus
Embodied AI
Current Work
Multi-Task Learning
Now / Focus
Now
Senior Staff Engineer at Huawei Singapore Norbert Wiener Research Centre. Building learning systems for long-horizon robotic tasks, with a focus on multi-task learning, world models, and scalable embodied intelligence.
Embodied AI
Multi-Task Learning
World Models
Long-Horizon Robotics
Focus
Advancing reasoning, generalization, and long-horizon control for embodied agents. I am interested in how learned world models and reusable skills can help robots plan, coordinate, and recover across diverse tasks.
Reinforcement Learning
Robotic Planning
Reasoning
Embodied Agents
Long-Horizon Decision-Making
Research agenda
Teaching robots to reason beyond the next action.
My work asks how embodied agents can transfer skills, remember what matters, coordinate with other robots, and recover when long-horizon plans meet an uncertain world.
Central question
How can one learning system turn experience into reusable intelligence?
I connect world models, multi-task reinforcement learning, and robotic planning to build agents that improve across tasks instead of solving each problem from scratch.
- 8
- Papers & preprints
- 6
- Applied systems
- 5
- Research directions
Recent Updates
Aug 2026
Role
Completed New Employee Orientation at Huawei Malaysia
Completed the New Employee Orientation programme at Huawei Malaysia in Kuala Lumpur.
Jul 2026
Publication
Dec-MARVEL Preprint Released
"Dec-MARVEL: Decentralized Multi-Agent Exploration without Communication under Budget Constraints" is on arXiv — joint work with Janghyun Cho and Changjoo Nam (Sogang University).
Jun 2026
Role
Started New Role at Huawei Singapore
Started as Senior Staff Engineer at Huawei Singapore Norbert Wiener Research Centre.
Jun 2026
Project
Vision-Guided Robotic Insertion for AIC
Built a vision-guided robotic system for fiber optic connector insertion, combining YOLO11n-pose, PnP pose estimation, multiview triangulation, and force-aware recovery.
May 2026
Publication
ORION Accepted to RA-L
"ORION: Option-Regularized Deep Reinforcement Learning for Cooperative Multi-Agent Online Navigation" has been accepted to IEEE Robotics and Automation Letters.
May 2026
Hackathon
Built Sonic Intelligence at AI Engineer Hackathon
Collaborated with Yan Rui Tan on Sonic Intelligence, a real-time audio AI system for environmental awareness, contextual sound understanding, and multimodal agentic pipelines.
Apr 2026
Workshop
OpenAI x Granite Asia Builder Workshop
Joined a Singapore builder sprint on OpenAI's latest developer stack, exploring rapid AI prototyping with Codex, voice interaction demos, and startup-focused AI workflows.
Apr 2026
Career
Huawei PhD Career Day at NUS
Attended the Huawei PhD Career Day NUS session, reflecting on research careers, real-world impact, and the many paths available after the PhD.
Mar 2026
Hackathon
Built NaviSwarm at Gemini 3 Singapore Hackathon
Participated in the Google DeepMind x Cerebral Valley hackathon and built NaviSwarm, a Gemini-powered indoor wayfinding assistant with multi-agent spatial memory, voice interaction, and AR guidance.
Jan 2026
Conference
AAAI 2026 at Singapore!
Attending AAAI 2026; collaborating with Joonyeol Sim (UCI) on Multi-Agent Path Finding for expanding maps!
Jan 2026
Event
Amazon AI Research Night
Engaged with leading scientists at the Amazon AI Research Night in Singapore.
Jan 2026
Event
Huawei Talent Night
Explored future tech horizons and networking at the Huawei Talent Night, Fullerton Bay Hotel.
Dec 2025
Conference
IEEE Multi-Robot Symposium (MRS) 2025 at Singapore
Attended IEEE MRS 2025; collaborating with Cho Janghyun (Sogang University) on distributed multi-robot exploration.
Aug 2025
Spotlight
Search-TTA Accepted to CoRL 2025
"Search-TTA" framework for visual search accepted as a Spotlight paper. Seoul, Korea.
June 2025
Publication
IEEE RA-L / IROS 2025
"Heterogeneous Multi-Robot Task Allocation" to be presented at IROS 2025 in Hangzhou, China.
May 2025
Presentation
ICRA 2025 Presentation
Presenting "MARVEL" in Atlanta, USA. Session Room 312 on May 22.
Apr 2025
Showcase
Invited Showcase @ SAFMC 2025
Live demo of outdoor evasive search using multiple drones in an urban environment mockup.
Mar 2025
Publication
IEEE RA-L Accepted
"Heterogeneous Multi-Robot Task Allocation" accepted to IEEE RA-L.
May 2024
Presentation
ICRA 2024 Live Demo
Presented live demo of swarm robotics algorithms at the ICRA 2024 Expo in Yokohama, Japan.
Apr 2023
Award
Judge Commendation @ SAFMC 2023
Awarded Judge Commendation for Category E: Swarm at Singapore Autonomous Flower Model Comparison (SAFMC) 2023.
July 2022
Scholarship
Awarded NUS-DSO Scholarship
Recipient of the prestigious NUS-DSO Graduate Program scholarship for Ph.D. research.
Aug 2019
Scholarship
NUS E-Scholar Scholarship
Awarded the NUS Engineering Scholarship (E-Scholar) for undergraduate studies.
Selected Research
MARVEL: Multi-Agent Reinforcement Learning for Constrained Field-of-View Multi-Robot Exploration in Large-Scale Environments
Introduces a decentralized MARL framework for scalable exploration for agents with constrained field-of-view.
STAR: Swarm Technology for Aerial Robotics Research
A comprehensive stack that simplifies the development and deployment of swarm algorithms on real hardware.
Search-TTA: A Multimodal Test-Time Adaptation Framework for Visual Search in the Wild
A framework that aligns remote sensing imagery with ground-level visual priors to improve robotic search efficiency using test-time adaptation.
Heterogeneous Multi-Robot Task Allocation and Scheduling via Reinforcement Learning
A hierarchical RL approach that jointly optimizes task scheduling and allocation for heterogeneous robot teams.
ViPER: Visibility-Based Pursuit-Evasion via Reinforcement Learning
Presents a learning-based solution for the visibility-based pursuit-evasion game. Unlike classical methods, ViPER utilizes a GNN-based policy.
Context-Aware Deep Reinforcement Learning for Autonomous Robotic Navigation in Unknown Area
A DRL approach for autonomous navigation that leverages context to improve performance in unknown environments.
3D Map Generation for Indoor Non-Manhattan World Environments
Research on generating 3D maps for complex indoor environments that do not follow the Manhattan world assumption and using a sparse LiDAR sensor.
Projects
Scene Reasoning
3D Scene Intelligence for Cross-Modal Foundational Search
This project develops a high-fidelity framework for embodied object navigation by leveraging incremental 3D Scene Graphs and foundational Vision-Language Models (VLMs). By moving beyond flat occupancy maps, the system builds a hierarchical semantic representation of the world that captures objects, rooms, and their relationships. The system utilizes Knowledge-Augmented Generation (KAG) to predict structural and semantic properties of unobserved regions, enabling robots to perform complex, cross-modal search missions based on categories, natural language descriptions, or visual exemplars in completely unknown topographies.
3D Scene Graphs
Vision-Language Models
Knowledge-Augmented Generation
View project
Embodied Navigation
Cognitive Architectures for Long-Horizon Semantic Navigation with Visual-Language Priors
This project explores the integration of foundational Vision-Language Models (VLMs) and Large Language Models (LLMs) to redefine the cognitive architecture of intelligent navigation. We investigated a framework that leverages the zero-shot reasoning capabilities of internet-scale foundational models alongside a structured spatial-semantic memory. This enables embodied agents to perform complex, language-driven semantic search tasks—such as finding specific objects described in everyday natural language—by reasoning over environmental uncertainty and past observations in completely novel environments.
Embodied AI
Long-Horizon Reasoning
Visual-Language Priors
View project
Multi-Agent Coordination
Scalable Decentralized Coordination for Complex Multi-Robot Systems
This project introduces MARVEL (Multi-Agent Reinforcement Learning for Constrained Field-of-View Multi-Robot Exploration), a framework for high-performance, decentralized coordination in large-scale environments. By leveraging Graph Attention mechanisms, MARVEL enables robot teams to reason about teammate intent and spatial dependencies under restricted sensing constraints. Our approach focuses on information-theoretic action pruning to optimize coverage and mission efficiency, facilitating complex collaborative maneuvers in completely unknown topographies without a central controller. Its follow-up, Dec-MARVEL, drops inter-robot communication altogether, coordinating on opportunistically observed teammate trajectories while holding every robot to a travel budget that guarantees a return to base.
MARL
Graph Neural Networks
Information-Theoretic Planning
View project
Search & Pursuit
Adversarial Multi-Agent Pursuit-Evasion in Complex Stochastic Environments
This project develops a decentralized, learning-based framework for visibility-based pursuit-evasion in challenging outdoor environments. We focus on enabling teams of mobile agents to systematically clear contaminated spaces and capture adversarial evaders within high-density urban terrains. By integrating multi-agent reinforcement learning with advanced spatial reasoning, the system addresses the critical challenges of building-induced occlusions and limited sensor ranges, allowing for real-time coordinated maneuvers without the need for a central controller.
Multi-Agent RL (MARL)
Multi-Agent Coordination
Graph Attention Networks
View project
Swarm Robotics
A Modular Infrastructure for Agile Nano-Quadcopter Swarm Autonomy
This project introduces STAR (Swarm Technology for Aerial Robotics), a modular, open-source infrastructure designed to bridge the gap between simulation and high-fidelity physical deployments. STAR integrates decentralized task allocation with robust vision-based landmark localization to manage fleets of nano-quadrotors (e.g., Crazyflies) in cluttered environments. The framework provides a high-throughput ROS 2-based communication layer and a hardware-in-the-loop (HIL) sim-to-real pipeline, enabling researchers to validate complex multi-agent algorithms, reactive obstacle avoidance, and swarm behaviors on physical robotic collectives.
Swarm Robotics
Sim-to-Real Deployment
Agile Autonomy
View project
Context-Aware Navigation
Leveraging context for robotic navigation in unknown environments
This project develops a decentralized framework for context-aware navigation, enabling embodied agents to perform complex path-finding and search tasks in unknown environments without prior maps. By leveraging Graph Attention Networks to encode environmental context, the framework allows robots to reason about the global structure of a space from local observations. This enables a suite of navigation capabilities—from zero-shot exploration to adaptive prior-based path-finding—that outperform traditional geometric planners in both computational efficiency and success rate.
Context Learning
Test-Time Adaptation
Multimodal Fusion
View project
Work Experience
Jun 2026-Present
Senior Staff Engineer
Huawei Singapore Norbert Wiener Research Centre
Multi-task learning and world models for long-horizon robotic tasks.
2022-Jun 2026
Associate Scientist
Temasek Laboratories @ NUS (Swarm Autonomy Lab)
Embodied AI research for multi-agent reinforcement learning, robotic navigation, and
coordinated robot autonomy.
Apr 2024-Jul 2024
Robotics Autonomy Intern
DSO National Laboratories (Robotics Division)
Academic Journey
2022-Present
2022
Bachelor of Engineering (Highest Distinction)
National University of Singapore
Jan 2021 – May 2021
Student Exchange Programme
Nanyang Technological University
Academic Service
Peer-review service
Contributing to the research community as a reviewer across robotics, AI, and automation venues.
AAAI2024, 2026
IEEE RA-L2025–2026
ICRA2024–2026
TASE2026
TMECH2026
ICCAS2024
IROS2024
CoRL2023
Honors & Awards
April 2025
Invited Showcase (Outdoor Evasive Search) @ SAFMC 2025
May 2024
Presented Live Demo (Swarm Robotics) @ ICRA 2024 Expo
April 2023
Awarded Judge Commendation @ SAFMC 2023 CAT E: Swarm
March 2023
Selected for Static Display @ SAFMC 2023
July 2022
Awarded NUS-DSO Scholarship under the NUS-DSO Graduate Program
July 2022
Attained Highest Distinction, Bachelor of Engineering, NUS
August 2019
Awarded the prestigious NUS E-Scholar Scholarship
Curriculum Vitae
AI Paper Reader
Don't have time to read a 20-page paper right now? Upload any research PDF to instantly generate a beautifully structured, comprehensive summary powered by Google Gemini.
Try the Tool