Skip to main content

Research Partnerships & Open Challenges

Duration: 35 min · Level: Foundational · Module: 10. G1 Development Roadmap · Focus: research, partnerships, challenges, strategy

Learning objectives

By the end of this lesson you will be able to explain and apply:

  • CMU Robotics Institute
  • MIT CSAIL & Biomimetic Robotics Lab
  • ETH Zurich RSL (Robotic Systems Lab)
  • UC Berkeley RAIL
  • Open challenges by 2030

Why this matters​

The humanoid robotics field is moving fastest through open collaboration.

Overview​

The humanoid robotics field is moving fastest through open collaboration. The labs and companies doing the most relevant work for G1 represent both potential partners and the knowledge base that will inform G1's development.

Key concepts​

Key idea

CMU Robotics Institute: whole-body control, dexterous manipulation, ALOHA bimanual systems — most directly applicable to G1 manipulation stack

  • MIT CSAIL & Biomimetic Robotics Lab: locomotion, actuator design, contact-rich manipulation — source of QDD actuation concept; strong collaboration history with startups
  • ETH Zurich RSL (Robotic Systems Lab): locomotion RL, ANYmal platform, sim-to-real — wrote most of the foundational papers G1's locomotion stack will rely on
  • UC Berkeley RAIL: Diffusion Policy, ALOHA, RT-2 collaborations — leading open-source VLA ecosystem; accessible through open-source code and Berkeley researchers joining companies
  • Open challenges by 2030: (1) long-horizon task planning (>50 step tasks), (2) robust outdoor locomotion in rain/wind/uneven terrain, (3) full 8-hour battery life with active manipulation
  • G1 data flywheel: every deployed robot generates real-world training data; 100 deployed G1s generating 10 hours/day each = 1000 hours/day → dramatically accelerates policy improvement

Check your understanding​

Try to recall each answer before expanding it.

Q1. What do you know about CMU Robotics Institute?

whole-body control, dexterous manipulation, ALOHA bimanual systems — most directly applicable to G1 manipulation stack

Q2. What do you know about MIT CSAIL & Biomimetic Robotics Lab?

locomotion, actuator design, contact-rich manipulation — source of QDD actuation concept; strong collaboration history with startups

Q3. What do you know about ETH Zurich RSL (Robotic Systems Lab)?

locomotion RL, ANYmal platform, sim-to-real — wrote most of the foundational papers G1's locomotion stack will rely on

Q4. What do you know about UC Berkeley RAIL?

Diffusion Policy, ALOHA, RT-2 collaborations — leading open-source VLA ecosystem; accessible through open-source code and Berkeley researchers joining companies

Q5. What do you know about Open challenges by 2030?

(1) long-horizon task planning (>50 step tasks), (2) robust outdoor locomotion in rain/wind/uneven terrain, (3) full 8-hour battery life with active manipulation

← Previous: 10.3 2026–2030 Development Milestones

Part of Module 10: G1 Development Roadmap.