The system view

WEAI research landscape

People set goals. Network-side intelligence coordinates across robots and shares task context. Each robot keeps its own perception, decision and control loop on board, so time-critical control does not wait for the network.

Between them sits the wireless link. Its bandwidth, delay and reliability decide what can be shared and how quickly — which is why we study communication, intelligence and physical action as one problem rather than three.

A representative system concept, not a deployment diagram. Joint-level control stays on the robot; the wireless link carries task-relevant information.

System concept: people set goals; network-side intelligence coordinates; task-relevant information crosses a wireless link subject to bandwidth, delay and reliability; each robot runs its own perception, decision and control loop; humanoid and quadruped platforms act in the physical world and return observations and task feedback.

Human goals & shared tasks

People set goals and work alongside robots

Goals & instructions

Progress & outcomes

Network side · edge & cloud

Network-side intelligence

Coordination, planning and shared task context

Wireless information exchange

Task-relevant information

Observations & evidence

  • Bandwidth
  • Delay
  • Reliability

On each robot

On-robot intelligence

  1. Perceive
  2. Decide
  3. Control

Local control loop — runs on board, without a network round trip

Actions

Physical observations & task feedback

Physical world

Humanoid & quadruped platforms

Acting on objects, spaces and people

Concept

From observation to verified action

The research loop that links our themes. Communication constraints appear where information is exchanged between the robot and the network.

Concept loop: observe, exchange information under bandwidth, delay and reliability constraints, build task context, select an action, execute it physically, verify the result, and observe again.

  1. Observe Sensors and robot state
  2. Task context Goal, progress and memory
  3. Action selection What to do next
  4. Physical execution Local control on the robot
  5. Result verification Did the action succeed?
  6. Then observe again — the loop continues.

Task-Grounded Robot Learning

We explore how embodied agents connect language, perception, task progress and physical action over extended tasks.

Focus areas

  • Vision–language–action models
  • Language grounding
  • Task progress
  • Memory
  • Long-horizon manipulation
  • Closed-loop verification and recovery

Projects in Task-Grounded Robot Learning

Ongoing research Task-Grounded Robot Learning

Progress-Aware Long-Horizon Robot Tasks

We investigate how robots can track task progress, maintain useful context and select subsequent actions during extended manipulation tasks.

Project investigator: Haonan Zhou

Ongoing research Task-Grounded Robot Learning

Language-Guided Closed-Loop Manipulation

We develop research prototypes that connect object perception, language instructions, grasp planning and feedback control for tabletop manipulation.

Project investigator: Haonan Zhou

Distributed Embodied Intelligence

We investigate coordination between network-side intelligence, on-robot models and heterogeneous robotic agents.

Focus areas

  • Network-side large models
  • On-robot models
  • Cloud–edge–robot task allocation
  • Multi-robot coordination
  • Shared state and evidence

Projects in Distributed Embodied Intelligence

Ongoing research Distributed Embodied Intelligence

Collaborative Intelligence Across Robots and Networks

We study how network-side models and on-robot intelligence can coordinate sensing, planning and execution across multiple robots, while accounting for communication and computation constraints.

Project investigator: Haonan Zhou

Ongoing research Whole-Body Control & InteractionDistributed Embodied Intelligence

Natural Interaction with Mobile and Humanoid Robots

We investigate interfaces that connect human instructions with robot behaviours, supporting interactive demonstrations across mobile and humanoid platforms.

Project investigator: Haonan Zhou

Wireless & Semantic Communication

We study how robots can exchange task-relevant information under bandwidth, latency and reliability constraints.

Focus areas

  • Task-relevant representations
  • Semantic communication
  • Compression and coding
  • Reliable transmission
  • Communication–control co-design

Projects in Wireless & Semantic Communication

Ongoing research Wireless & Semantic Communication

Task-Aware Communication for Embodied Intelligence

How much information does a robot need to act? We investigate compact representations and communication strategies that preserve task-relevant information across constrained and unreliable links.

Project investigator: Haonan Zhou

Whole-Body Control & Interaction

We investigate movement, manipulation and interaction while accounting for whole-body dynamics and physical constraints.

Focus areas

  • World models
  • Whole-body dynamics
  • Loco-manipulation
  • Balance
  • Geometric constraints
  • Human–robot interaction

Projects in Whole-Body Control & Interaction

Ongoing research Whole-Body Control & Interaction

Balance-Aware Whole-Body Manipulation

We explore how whole-body dynamics and geometric constraints can inform upper-body motion, with the aim of improving stability during humanoid manipulation.

Project investigator: Haonan Zhou

Ongoing research Whole-Body Control & InteractionDistributed Embodied Intelligence

Natural Interaction with Mobile and Humanoid Robots

We investigate interfaces that connect human instructions with robot behaviours, supporting interactive demonstrations across mobile and humanoid platforms.

Project investigator: Haonan Zhou