Sunday Robotics: Scaling the Home Robot Revolution with Co-Founders Tony Zhao and Cheng Chi

What It Covers

Sunday Robotics co-founders Tony Zhao and Cheng Chi explain how they're building Memo, the first general home robot, using scaled imitation learning with 10 million trajectories collected via custom gloves to achieve dexterous manipulation.

Key Questions Answered

• •Data Collection Innovation: Sunday developed three-finger printed gloves with GoPro tracking that enables 500 people to collect robotic training data in real homes without robots present, achieving 10 million trajectories versus traditional lab-only teleoperation setups that require PhD-level expertise.
• •Hardware Design Philosophy: The robot uses compliant, low-cost actuators instead of precise industrial ones because AI vision allows real-time correction of mechanical inaccuracies. Three fingers replace five by combining naturally-grouped digits, reducing cost threefold while maintaining 95 percent of manipulation capability for home tasks.
• •Scaling Recipe Discovery: Training stability emerged only after 20 glove iterations and extensive data filtering pipelines. Quality control automation detects hardware failures before shipping, and the team learned that data quality matters more than quantity when scaling from thousands to millions of trajectories in diverse environments.
• •Product Timeline and Economics: Beta program launches in 2026 with robots in customer homes. Manufacturing cost drops from $20,000 to under $10,000 at scale through injection molding versus CNC machining. Commercial availability targets 2027-2028, not a decade away, contingent on beta reliability validation.