''Physical Intermediary of Robots Absolutely Necessary to Solve Real-World Problems''
Dennis Hong (UCLA Robotics & Mechanisms Laboratory) presented at AI Expo MARS 2025 opening ceremony on urban AI and robot collaboration. Key insight: instead of asking "Are humanoid robots necessary?" the question should be "What tasks require human-shaped robots?" — form follows function (Louis Sullivan). The case for humanoid form: human environments (stairs, doors, furniture) are all designed for human body dimensions; robots coexisting with humans in these environments need compatible physical form. Lab achievements: "Arly Vis" — robot that doesn''t fall when forcefully pushed or kicked; "first running robot" — both legs simultaneously leave the ground; robot autonomously navigated the 4th floor of the laboratory using AI (not remote control); helium balloon leg robot demonstrating graceful walking ("the world''s safest robot"). Extreme terrain capability: robot navigates grass, gravel, and asphalt surfaces autonomously; uses contact sensors in feet (not cameras) to navigate 5cm obstacles — a critical advance for real-world deployment where visual sensing alone is insufficient. City intelligence thesis: Urban AI requires robots as physical intermediaries because most valuable urban problems are physical — emergency response, infrastructure inspection, elderly care, last-mile delivery. AI without physical embodiment can optimize routing and predict failures but cannot physically intervene when intervention is needed. The integration challenge: making AI and robot systems that individually work well also work together in real urban environments requires solving coordination, communication, and shared situational awareness problems that neither AI nor robotics research alone addresses — which is why Hong argues their integration is the key technical challenge for urban AI to become genuinely functional.
