Skip to content

Research Directions and Project Opportunities

Public-information boundary

This page provides only laboratory-approved direction-level summaries. Participant lists, current allocations, selection results, unpublished experimental data, internal roadmaps, and recruitment contacts are not published in the public Wiki.

Purpose

Help laboratory members understand current robot-learning directions and find the appropriate controlled contact channel.

Direction Overview

DexGello

A data-collection and teleoperation direction for dexterous manipulation, exploring high-quality action mapping among dexterous hands, isomorphic robot arms, and human demonstrations, together with integrated collection, training, evaluation, and deployment workflows.

OmniEgo

A human-intent-oriented multi-sensor data-collection direction studying how VR, vision, wearable sensors, and environmental sensors can jointly represent human actions and task context.

RoboDojo-Tac

A visual-tactile evaluation direction for contact-rich manipulation, focusing on task definitions, sensor fusion, reproducible evaluation, and sim-to-real alignment across real and simulated environments.

Robot4Robot

A direction studying how embodied agents and external agents collaborate through multimodal sensors, spanning real robots, simulation modeling, 3D scenes, world models, imitation learning, continual learning, and self-improvement.

  • Vision-language-guided robotic manipulation
  • Learning dexterous manipulation from human demonstrations
  • Visual-tactile fusion for contact-rich manipulation
  • Simulators, world models, and world action models
  • Multi-robot and multi-sensor collaboration
  • Robot data collection, evaluation, and closed-loop self-improvement

Participation Process

  1. Read the direction summaries and confirm course, time, and on-site experiment commitments.
  2. Contact the Project Coordinator through the controlled directory with your background, interests, and availability.
  3. Confirm scope, required skills, data boundaries, and expected outcomes with the project owner.
  4. If course credit is required, complete the separate eligibility, approval-letter, and official registration process.
  5. Add members to controlled project documents and repositories only after owner approval.

Information Maintenance Rules

  • Recruitment state, capacity, and participant allocation are dynamic and are not fixed on this page.
  • A project owner must approve the name, summary, and disclosure boundary before public presentation.
  • Never upload signup chains, group-chat screenshots, participant lists, or spreadsheets containing personal information.
  • Review project status at least once per term; mark unconfirmed information as “Requires maintainer review.”

Maintenance

  • Owner: Research Project Coordinator
  • Contact: <CONTROLLED_CONTACT_DIRECTORY_URL>
  • Last verified: 2026-08-03