Sunday's Vision-First Robotics Strategy

This design choice is really a bet that software can be cheaper to improve than hardware. Traditional robots pay for accuracy in metal, with expensive motors, gearboxes, and calibration so the arm lands in the exact spot every time. Sunday instead lets the arm be a little sloppy, watches the scene through cameras, sees where the hand and object actually are, and corrects the motion on the fly, which pushes cost out of parts and into models that can keep getting better after the robot ships.

• The broader home robotics split is between humanoids like 1X and Figure, which use more complex bodies to handle stairs, cabinets, and many chores, and simpler non humanoids like Sunday, which trade task breadth for a lower bill of materials and a sub $10K target price.
  1 sacra 4 sacra
• This approach already has a close consumer precedent in Matic. Matic uses camera based perception and on device neural networks to map a home and clean floors without leaning on traditional lidar heavy navigation, showing how better vision can let a robot replace costlier sensing or mechanical precision with software.
  5 sacra
• The catch is that vision corrected robotics only works if the model has seen enough messy real world edge cases. Sunday is pairing cheaper hardware with glove based household demonstrations across 500 plus homes, because the data has to teach the robot how to recover when the arm is off by an inch, the toy is rotated, or the room layout is unfamiliar.
  1 sacra 2 sacra

If this works, home robots will start to look more like smartphones than factory machines. The winners will be the companies that can ship affordable hardware early, gather the most household interaction data, and steadily improve behavior through updates, rather than the companies that try to engineer perfect mechanics upfront.