In the span of a year, we have witnessed the death of major consumer robot companies. What went wrong? Guy Hoffman, a robotics expert, provides a comprehensive summary of possible reasons for the demise of the companies. To me, the problem was good old over-promise and under-deliver. I mean, just look at this commercial and look at a real robot. But I also think that there is another problem that is rooted in our culture; when you hear the word “robot”, what comes into your mind? I think about C-3PO and R2-D2 from Star Wars, The Terminator, WALL-E, Sonny from I, Robot, and Marvin from The Hitchhiker’s Guide to the Galaxy, just to name a few. Most of these robots are physically more capable than humans and even emotionally as capable as humans[^1]. After seeing such robots over and over, we expect a robot to be a super awesome friend/servant (or killing machine), and see a commercial like this and we get this[^2]. But I digress.

Of course not all consumer robots are dead. We still have iRobot’s Roomba and Amazon’s Echo/Alexa–if you consider a voice-agent compatible smart speaker a robot. But for some reason, they don’t feel like a robot. I nearly gave up on trying to define what the “robot” is and advocating for not calling anything a “robot”, partially to stop that cultural image of the “robot” mentioned earlier. But that didn’t work, e.g., I couldn’t stop using the word “robot”. So I made peace with considering a physical device (e.g., a mobile robot or robot manipulator) that is capable of complex sensing (e.g., computer vision), complex control (e.g., motion planning), and adaptation (e.g., machine learning) a “robot”. For example, I wouldn’t call industrial manipulators that pick things up from and place things in known locations but I would call a mobile rover that recognizes street signs and avoids pedestrians crossing a road a robot.

Okay, so where do we go from here? Will we have a “robot” that we can use at home? what will it look like? what will it do? It’s likely that a big tech company will build something, but I say the time is now for rolling our sleeves up and building home robots ourselves! It might be difficult but I think it is possible. First, we need hardware. We can buy a kit like a TurtleBot or get an open hardware design from one of the Hackaday blog posts or design one from scratch on onshape, if you can. Second, we’ll need to add electronics. We can buy a Raspberry Pi or a super fancy GPU or TPU board from NVidia or Google. I can’t work with microcontrollers but don’t let me stop you. Finally, software. There are opensource softwares like ROS, Nvidia Isaac, Apex Autoware, etc.

What should we build?[^3] That’s a great question. My approach is reviewing my hobbies and asking how can I use robotics capabilities to make them more interesting. This usually boils down to adding capabilities like mobility, manipulability, or computer vision to existing things. I tried to build interactive or mobile decorations like interactive lights (e.g., changing ceiling light colors based on locations of detected people) and mobile pots (e.g., moving to different locations at different times). Another approach is starting from the problem, e.g., by asking what problem can I solve with robotics capabilities? This approach usually gets blocked by the current limitations of the robotics tech, but some people do come with clear solutions (e.g., limiting the scope of the problem, involving humans, etc.) to make progress.

Honestly though, it isn’t easy to build homemade robots and making them actually useful is near-impossible for a hobbyist. Although robotics technology is advancing quickly, there are limitations that makes them not as reliable for real world use cases and hardware devices (e.g., high-quality motors) are still pretty expensive. So why did I suggest to build your own robot? I wanted the robot lovers to acknowledge the fact that we are building home robots because we just love doing so[^4] and you shouldn’t be discouraged by recent fallings of the robot companies. Long live DIY robots!


Updates

Footnotes

[^1] I love fictions/movies like Pluto or Her that questions the meaning of “human” when robots can be as capable as humans.
[^2] I keep picking on Jibo but I really wanted it to succeed and at this point, I think learning from its mistake is important for roboticists of tomorrow.
[^3] Also checkout How to Pick a Problem
[^4] Checkout AN APP CAN BE A HOME-COOKED MEAL by Robin Sloan