Robot Pets

Robots are all the hype these days.

Almost every robotics company today is struggling with figuring out how it can be useful. Companies like Figure and Boston Dynamics have yet to make profit (the backflips were cool though, rip Atlas). They’re propped up by a constant influx of investment on the future promise of generating extreme wealth. This is almost undoubtedly true, but it’s also what the AV industry has been promising for the last 15 years. Perhaps this time is different, but history has a tendency to repeat itself.

The question is what do you use these robots for and who’s going to pay for it? So far, the only good answer we’ve got are industrial robots (and maybe robot vacuums), but they’re confined to fixed tasks and environments. Fundamentally, it’s just really hard to beat human labour with robot labour. Humans are extremely dexterous and capable, and relatively cheap all things considered. We’ll get there eventually (just like how I think autonomous cars will eventually get there), but it’s going to be a long, difficult, expensive road.

But replacing human labour is not the only way to sell robots. Disney has released some very impressive demos of some of their character-based robots. On that demo video, one of the comments got me thinking.

The appealing nature of Disney’s robots are the personalities embodied within them. It’s the same reason we fall in love with the quirkiness of C3P0 and find R2D2 irresistibly cute. K9 from Dr. Who anyone? Or Wall-E. Everyone loves Wall-E. I don’t think it’s a stretch to say that it’s also one of the biggest reasons people buy pets, like dogs. And the pet market is huge ($250 billion).

The reason this category of robots holds a particular fascination to me is for a couple of reasons (Lex Fridman has a great podcast episode with Boris Sofman where they talk about robot toys):

1. Consumers are much more forgiving of mistakes. Maybe your robot falls over occasionally. For a little robot that waddles around your house? Cute!
2. You don’t need expensive hardware to make a lovable robot (case and point). Though to have potential of being useful and yet still affordable is a huge engineering challenge. But we’re not far off. The cheapest Unitree robot dogs are roughly one order of magnitude off the hundreds of dollars mark that I would imagine a robot like this would need to cost in order to be feasible. Over time we may yet bring costs down further.
3. People spend hours and hours training their pets to do various tricks. Will people do that with their new robots? Can you teach your robot to take out the trash? Can you create an unparalleled distributed training system for your robots that out scales any company on earth?

Perhaps a revolution of personality is needed to get the adoption you need to train the capabilities that you want. Maybe we embrace the uselessness of robots so that one day, we may develop ones that are.