Cars are essentially wheeled robots. Drawing parallels between cars and humanoid robots reveals important insights about the path to full autonomy.
Traditional cars were designed from the start to be operated by humans. In contrast, the long-term vision for humanoid robots is complete autonomy. Yet the development of autonomous vehicles shows a critical pattern: self-driving cars were trained extensively on real-world data from human drivers. This data taught AI systems how to navigate alongside human drivers, handle diverse scenarios, and align with human behaviors and preferences. Only after proving superior performance could human input be safely removed.
A similar progression is highly relevant for humanoid robots. An intermediate stage of human teleoperation can generate rich, high-fidelity training data on how robots should behave in real-world tasks and environments. It captures human expectations, decision-making, and natural interaction styles — enabling robots to learn context-aware, socially appropriate responses.
This means that widespread human-operated humanoid robots will likely precede fully autonomous ones in real-world deployments. Just as human-driven cars created the foundation for today’s autonomous vehicle revolution, human-in-the-loop teleoperation will accelerate the safe and effective integration of humanoids into society.
Why Locomotion Control Matters
A key operational difference highlights why specialized tools matter. Driving a car primarily involves locomotion control, elegantly handled by pedals and steering wheel. Operating a humanoid robot, however, requires simultaneous mastery of both locomotion (navigating dynamic environments) and sophisticated action control (dexterous hand/arm tasks essential for real-world usefulness).
Foottroller serves as the natural “pedals and steering wheel” for humanoid teleoperation. By enabling intuitive, foot-based locomotion control, it frees the operator’s hands for precise, complex actions — making human teleoperation far more effective and scalable for generating the training data needed to reach full autonomy.
Ready to bridge the gap between human skill and robotic autonomy?