199 lines
6.7 KiB
Markdown
199 lines
6.7 KiB
Markdown
# Humanoid Robotics Context
|
|
|
|
**Created:** 2026-03-11
|
|
**Status:** Draft
|
|
**Related:** [[Hexagon Role Analysis]], [[Job Hunt]], [[Career]]
|
|
|
|
---
|
|
|
|
## Why Humanoid Robotics Matters Now
|
|
|
|
Humanoid robots are having a moment. After decades of research and failed commercial attempts, several factors have converged to make humanoids viable:
|
|
|
|
1. **Hardware costs dropped** - Actuators, sensors, compute have become affordable
|
|
2. **AI/ML matured** - LLMs enable natural language interfaces; vision systems are robust
|
|
3. **Labor shortages** - Aging populations in developed nations create demand for automation
|
|
4. **Investment flood** - Billions in VC and corporate funding since 2022
|
|
|
|
This isn't speculative anymore. Real companies are deploying real robots in warehouses, factories, and soon homes.
|
|
|
|
---
|
|
|
|
## Key Players (2026)
|
|
|
|
### Tier 1: Serious Contenders
|
|
|
|
| Company | Focus | Funding/Backing | Status |
|
|
|---------|-------|-----------------|--------|
|
|
| **Tesla Optimus** | General-purpose, mass production | Tesla internal | Aggressive timeline, targeting 2025 production |
|
|
| **Figure AI** | General-purpose, enterprise | $675M (2024) + OpenAI partnership | Pilots with BMW, other manufacturers |
|
|
| **Agility Robotics** | Warehouse/logistics | $150M+ | Digit in production, deployments with Amazon |
|
|
| **Boston Dynamics** | Atlas platform | Hyundai | Advanced R&D, shifting to production |
|
|
| **1X Technologies** | Security/home | $100M+ | NEO robot in pilot deployments |
|
|
|
|
### Tier 2: Emerging / Specialized
|
|
|
|
| Company | Focus | Notes |
|
|
|---------|-------|-------|
|
|
| **Apptronik** | Industrial humanoid | NASA heritage, Apollo robot |
|
|
| **Sanctuary AI** | Cognitive robots | Phoenix robot, Canada-based |
|
|
| **Unitree** | Low-cost humanoids | Chinese, H1 model ~$100K |
|
|
| **Fourier Intelligence** | Rehab robotics | Chinese, GR-1 humanoid |
|
|
| **Pal Robotics** | Research platforms | Spanish, TIAGo family |
|
|
|
|
### Corporate Initiatives
|
|
|
|
| Company | Notes |
|
|
|---------|-------|
|
|
| **Hexagon Robotics** | New division (2025), Zurich-based, precision heritage |
|
|
| **Toyota Research Institute** | Long-term research, robotics focus |
|
|
| **Honda** | ASIMO heritage, renewed interest |
|
|
| **Samsung** | Investing in robotics startups |
|
|
|
|
---
|
|
|
|
## Technical Landscape
|
|
|
|
### Core Challenges (Still Unsolved)
|
|
|
|
1. **Locomotion** - Bipedal walking on uneven terrain, dynamic balance, recovery from pushes
|
|
2. **Manipulation** - Dexterous hands, contact-rich manipulation, tool use
|
|
3. **Perception** - Real-time scene understanding, object recognition, human intent
|
|
4. **Planning** - Whole-body motion planning, task-level planning, real-time execution
|
|
5. **Control** - MPC for underactuated systems, contact dynamics, safety
|
|
6. **Integration** - Bringing it all together in real-time
|
|
|
|
### Why These Challenges Are Hard
|
|
|
|
- **Underactuation:** Humanoids have fewer actuators than degrees of freedom
|
|
- **Contact dynamics:** Contact forces are discontinuous, hard to model
|
|
- **Real-time constraints:** Planning/control must run at 100+ Hz
|
|
- **Safety:** Robots near humans require fail-safety
|
|
|
|
### State of the Art
|
|
|
|
**Locomotion:**
|
|
- MPC-based walking controllers (Boston Dynamics, IHMC)
|
|
- Learning-based approaches (reinforcement learning for robustness)
|
|
- Hybrid dynamics modeling
|
|
|
|
**Manipulation:**
|
|
- Contact-implicit planning
|
|
- Learning from demonstration
|
|
- Tactile sensing integration
|
|
|
|
**Planning:**
|
|
- Whole-body motion planning (still research-level)
|
|
- Behavior trees for task orchestration
|
|
- Integrated perception-planning-control
|
|
|
|
---
|
|
|
|
## Market & Career Implications
|
|
|
|
### Why Work in Humanoids Now?
|
|
|
|
1. **Timing:** Industry is transitioning from research to early deployment
|
|
2. **Skills:** Robotics generalists with deep expertise in one area are valuable
|
|
3. **Network:** Early players will shape the industry for decades
|
|
4. **Mobility:** Skills transfer across robotics domains
|
|
|
|
### Career Paths
|
|
|
|
**Technical IC:**
|
|
- Senior Engineer → Staff Engineer → Principal Engineer
|
|
- Deep expertise in planning, control, perception, or manipulation
|
|
- Can stay technical indefinitely if desired
|
|
|
|
**Technical Leadership:**
|
|
- Senior Engineer → Tech Lead → Engineering Manager → Director
|
|
- Requires cross-team coordination and people skills
|
|
- System architects often move into leadership
|
|
|
|
**Founding / Early-stage:**
|
|
- Join early-stage company, get equity
|
|
- High risk, high reward
|
|
- Best if you want ownership and can tolerate uncertainty
|
|
|
|
### Skills Premium
|
|
|
|
**High-value skills:**
|
|
- Whole-body motion planning
|
|
- MPC for underactuated systems
|
|
- Contact-rich manipulation
|
|
- Real-time control systems
|
|
- Integration architecture
|
|
|
|
**Commoditizing skills:**
|
|
- Basic ROS proficiency
|
|
- Simulation setup
|
|
- General Python scripting
|
|
|
|
---
|
|
|
|
## Hexagon Robotics Position
|
|
|
|
**What we know:**
|
|
- New division (2025) of Hexagon AB
|
|
- Based in Zurich
|
|
- Developing humanoid robots
|
|
- Parent company is global leader in precision measurement and metrology
|
|
|
|
**Implications:**
|
|
- **Stability:** Backed by large, profitable parent (less startup risk)
|
|
- **Precision focus:** Hexagon's heritage suggests emphasis on precision manipulation
|
|
- **Industrial applications:** Likely targeting manufacturing, quality control
|
|
- **Zurich ecosystem:** Strong robotics cluster (ETH Zurich, multiple companies)
|
|
|
|
**Career assessment:**
|
|
- Early-stage opportunity with growth potential
|
|
- Less equity upside than VC-backed startups, but more stability
|
|
- Technical work likely focuses on precision manipulation and industrial applications
|
|
- Good location for Claudio (Zurich-based)
|
|
|
|
---
|
|
|
|
## Learning Resources
|
|
|
|
**Locomotion:**
|
|
- "Legged Robots That Balance" - Marc Raibert (classic)
|
|
- IHMC walking controller papers
|
|
- Boston Dynamics Atlas papers/videos
|
|
|
|
**Whole-body planning:**
|
|
- "Whole-Body Motion Planning" literature (ROS, IROS conferences)
|
|
- Contact-implicit trajectory optimization papers
|
|
|
|
**Manipulation:**
|
|
- "Robotic Manipulation" course (CMU, Stanford online)
|
|
- Dexterous manipulation literature
|
|
|
|
**General:**
|
|
- RSS (Robotics: Science and Systems) conference
|
|
- IROS, ICRA conferences
|
|
- Robohub.org for industry news
|
|
|
|
---
|
|
|
|
## Open Questions for Hexagon Application
|
|
|
|
1. What's the team size? (Early = more ownership, later = more structure)
|
|
2. What's the technical stack? (ROS2? Custom? Simulation tools?)
|
|
3. What's the application domain? (Industrial? Service? General-purpose?)
|
|
4. How does Hexagon's metrology heritage influence the robot design?
|
|
5. What's the trajectory? (Research prototype → product timeline?)
|
|
6. Team culture? (Research-heavy? Engineering-heavy? Product-driven?)
|
|
|
|
---
|
|
|
|
## Notes
|
|
|
|
- Industry is real this time, not just hype
|
|
- Technical challenges are hard but solvable
|
|
- Career timing is good: early deployment phase
|
|
- Hexagon = stable parent + early-stage opportunity
|
|
- Focus on motion planning and integration roles matches market demand
|
|
|
|
---
|
|
|
|
**Tags:** #humanoid-robotics #industry-context #career #hexagon |