Back to Project List
RecruitingAIResearch

Locomotion Development with a Quadruped Robot

Using Unitree Go2 R&D, this project develops locomotion that smoothly traverses steps.

Members: Several
Started: 2026-08

Project Details

Using a loaned Unitree Go2 R&D platform, this project develops and validates locomotion methods that can smoothly traverse steps and uneven terrain on real hardware.

Objective

Build robust step-traversal locomotion by applying Physical AI methods centered on reinforcement learning and bridge the results to real-robot deployment.

Location

International Science Innovation Building, Yoshida Campus, Kyoto University

※ Unitree Go2 R&D is loaned by Takebishi Corporation.

Implementation

The team will combine simulation-driven development and real-world evaluation to improve locomotion robustness over obstacles.

Key Initiatives

  • -Locomotion model development using Unitree Go2 R&D
  • -Application and validation of reinforcement-learning-centered methods
  • -Use of simulators such as Isaac Sim
  • -Considered use of LiDAR and depth cameras
  • -Planned foundational lectures for members new to this area

Publication Policy

Deliverables will be published under Apache License 2.0, CC BY 4.0, or CDLA Permissive 2.0. Intellectual property rights remain with participants.

Open Roles and Requirements

Open Positions

Several

Schedule

Recruitment: Until 8/14 (may be extended depending on circumstances)

Implementation period: 2026/08/05-2026/12/31

Required

  • -Primarily Kyoto University students at doctoral level or below
  • -Students from other universities can be considered upon consultation

Preferred

  • -AI development experience in Python (especially RL and Sim2Real)
  • -Linux development experience
  • -Team development experience

Foundational instruction is planned, assuming some members may be new to reinforcement-learning-centered Physical AI development.

Benefits

  • -The loaned Unitree Go2 R&D can be used within project scope.
  • -Compute resources such as GPUs required for development are provided.
  • -Deliverables can be listed in portfolios, and publication credits are available on request.

How to Apply / Contact

Please send a message on KUPAC Discord with your affiliation (school/department) and your intention to join.