AI summary by TwinMind
Next Up for AI? Dancing Robots | Catie Cuan | TED
TED
10m 15s
Speaker's Background and Unusual Career
- Speaker transitioned from human stage dancing to robot choreography in 2017
- Works with diverse robot types: big, small, single arms, roving, flying, vacuuming robots
- Some humanoid, many non-humanoid designs
- Claims to have choreographed/danced with more robot types than anyone on Earth
- Also roboticist and engineer studying robot motion-emotion connection
- Works in emerging field of "Cho robotics" (intersection of Dance and Robotics)
Core Philosophy: Movement Affects Human Perception
- Robot movement directly impacts human emotional response
- Polite sliding through doorway = feeling seen/acknowledged
- Quick marching toward person with last-second avoidance = revulsion/fear
- Robots appearing in everyday environments: sidewalks, offices, backyards, hospitals
- Will be threatening/confusing without careful movement examination
- Roboticists focus on efficiency, speed, success optimization - insufficient approach
The Case for Teaching Robots to Dance
- Dancing robots demonstrate: intention, agility, balance, emotional expression, grace
- Will create better caregivers, teachers, companions
- Enables building general purpose robots people want to live with
- Expands future possibilities for intelligent machines
- Choreographers understand how movement affects humans
Historical Context and Evolution
- Robot-choreographer collaborations since 1990s
- Highly scripted performances captured public imagination
- Created wonder, laughter, desire to dance, sometimes fear
- No longer just art/entertainment - transferring movement lessons to everyday robots
- Two key lessons: 1) movement changes emotions, 2) textual movement critically important
AI Revolution in Robot Movement
- AI giving robots physical intelligence through robot bodies
- GPT and Gemini becoming robot brains - robots getting much smarter
- AI impact on text/video generation just scratching surface for physical world
- Before AI: hours needed to script simple dance sequences
- With AI: teach robot few specific doors, learns to open all doors (even unseen)
- Same principle for dance: teach specific person interaction, learns to dance with many others in different environments
Project Starling: Google's Everyday Robots
- Speaker worked at Everyday Robots (Google AI moonshot)
- Used AI to teach 15 robots moving together as flock
- Vision: robots parting like doves/crowd on city street for humans
- Goal: seamless, beautiful navigation through busy chaotic environments (Times Square example)
- World too complicated for step-by-step scripting/choreography
- Required AI magic for improvisation and reaction
- Massive unprecedented challenge, 2 years hardcore engineering
Project Starling Technical Implementation
- Trained AI agent on choreographer preferences (speaker's choices)
- Selected most engaging patterns and sequences
- Taught robots human gesture recognition and response
- Converted robots into musical instruments
- Each joint mapped to different sound
- Torso movement = bass sound
- Hand opening/closing = bell sound
- Robot actions created unique symphony
- Result: interactive performance and installation work
Project Starling Experience and Impact
- AI-generated movement with robot-generated music
- Designed to feel like stepping from normal reality into waking dream
- Far from current reality yet felt like safe world for children
- Natural, safe feeling despite machine surroundings
- People initially tentative, then engaged with curiosity/playfulness
- People kept returning to experience
Future Applications: Senior Living Scenario
- Robot in senior living facility crowded café
- Human sits, waves, makes sipping gesture
- Robot understands water request
- Robot drives over while nodding at people, pausing graciously
- Places glass gracefully, nods acknowledgment when thanked
- Possibly performs small celebratory dance
- Demonstrates gesture navigation, human interaction, expression applicable to general robots
Cultural and Historical Robot References
- Promise of coexisting robots existed long time
- Karel Čapek's 1920 play "Rossum's Universal Robots"
- Jetson's Rosie the robot
- Star Wars C-3PO
- Terminator
- Stories deeply impactful on imagination and building
Vision for Robot Capabilities
- Beyond chores: robots as dance partners
- Humans do diverse activities: ceramics, diplomacy, choir singing, vertical farming
- Robots should do myriad functional and creative things
- General purpose robots addressing demographic challenges
- Aging populations and massive labor shortages
- Movement quality determines home/workplace acceptance
- Maintaining human marvel in increasingly AI world
Conclusion: Power to Choreograph the Future
- Humans have power to choreograph desired future
- Movement design critical for robot acceptance
- Choreographic approach enables welcoming, understood robots people want to spend time with
