Bespoke Robot Society:Beyond Toys

Beyond Toys

This is BRS rant number two, exploring the distinction between toys and robots, and how we can use toys as a starting point without making them dead ends.

The Question

How do we distinguish between a robot and a toy? Especially when the robots are mostly being advertised toward or made accessible to children, there's actually going to have a lot of overlap with a toy.

But we want to move beyond toys.

Work vs Play

While exploration is a great form of play, not every toy enables you to learn something from it. Not every toy is extensible. A lot of toys just make some kind of amusement available, but they don't make the mechanics of the toy itself available. A lot of toys cover up their internals as a means of safety or protecting intellectual property.

Both of those causes are not applicable to Bespoke Robot Society's robots.

Toys are for play. Robots are for work. And sometimes the work we want to do is learning and teaching.

See also: BattleBots Are Not Robots for more on what makes something a real robot versus a remote-control toy.

Toys vs Tools

Toys, like many beginner robotics kits, are designed with limited complexity and predefined goals—like following a line, bumping into walls. If all the thing does is that single task that it's built for, it could technically be a robot, but you're still probably looking at a toy.

A true robot, on the other hand, is a tool designed to interact with an environment.

Forward-Looking Robots

I want to have a forward-looking definition of robot in Bespoke Robot Society.

The old definition of robot would be those giant arms that assemble cars. They move very precisely, they have a lot of strength, they are enormously stiff, and you cannot go near them while they're in motion for your own safety. They're uncompromising. They are executing and repeating some motion a lot. Many of these robots may not have real feedback loops in the sense that they aren't really able to detect if their task is going well or not. They may require a human operator to observe and hit the e-stop when they do something faulty.

But a forward-looking definition of robot is about autonomy. It's about reacting to your environment without planning it out perfectly ahead of time.

What I mean by that is: if we have goals and we have capabilities and we have sensors, but we don't necessarily know everything about the environment—people could walk by, you don't need to bump into them, the terrain may have changed, stuff could move—we want to use our sensors and figure that out dynamically. That's a real robot.

The Problem with Staying a Toy

Specifically, to advance away from a toy: a non-toy robot enables you to understand the problem it tries to solve and understand the limitations of the environment it can work in.

Then if you get a new constraint, a new difficulty added to that environment, or you want to move the robot to a new environment, if you understand how the robot works and how it relies on those environmental limitations, you can enhance the elements of the robot that don't suit the changes to the environment and you'll be able to create that new functionality yourself.

That's really the goal: enabling real-world use cases. What we mean is, you think of something that your robot might be able to do, and now you need to:

1. Understand the limitations it's currently built around
2. Have some resources about how to go beyond those current limitations and make the robot more robust, or robust against different hazards

Staying in the toy phase has an allure of false mastery. It's like playing with a toy that you understand fully. You might understand what, but not why. You may fully know how to operate, but you don't know how to extend. You may get proficiency as a user, but you will not become a creator.

That's one of the main problems with toys: even when Bespoke Robot Society robots make good toys, I don't want them to remain toys or be limited by the definition of toy.

Tech Trees

One of the ways I want to address this is with tech trees.

Bespoke Robot Society's $20 robot like SimpleBot might come in a kit with every single part. The PCB is already made, and you have no real requirements put upon you besides using a Phillips head screwdriver. Now you can hit go, and you can see the robot do the task it was immediately built for.

That's great because we want people to feel the confidence, we want them to get the rapid feedback of being able to play. We want them to have a working thing quickly, not a pile of parts or an endless project.

But where do you go from there?

From Kit to Custom

You could mod your robot, you could add new parts, you could repeat the robot but from a lower level of pre-completion.

Maybe you get another kit which does not include any 3D-printed parts. You go to the library and you 3D-print the parts that you need. Now you get to pick the colors, you might have to deal with a print failure, you have to think about the filaments you're going to use.

Or maybe instead of a pre-built PCB, you can solder together a kit. It's the same circuit that you're familiar with, but it's just through-hole soldering. Or maybe you want to learn the surface-mount version of that board, so you go and get the smaller SMD components.

Now we start to establish that equivalence: the prefab one is a factory-made SMD board, but the one you put together yourself is a through-hole board. But all the components are equivalent, the behavior is equivalent.

Or we go all the way for artisanal, and we get a protoboard and we hand-wire it. Again, it could be equivalent, but now you're looking at something that's not just a circuit board—it is a steampunk work of art. It is a retro-aesthetic design choice, not just a part that completes the requirements or does the behavior. It's an expression of your mastery that you've gone forward to do that.

Same thing with the software. If you have a robot that does some task, you can improve it just by changing the software. The same hardware everyone has—whether it's surface-mount, through-hole, self-made, completely custom—if it meets the standard, if it does the same task, we can share that same software. Now you're not just learning about how to change things, you're also collaborating with other folks.

This relates directly to giving robots away for free—when the designs are open and shared, everyone benefits from improvements.

The Starting Point

There's this huge tech tree, and we want everyone to be able to hop on and get the lay of the land by using pre-built robots, kits that are mostly assembled, things like that. But we also want you to be able to use your perspective of having that working robot to see all the different paths you could go down.

Toys are a starting point—we want you to turn it on and play immediately. But toys cannot be the end, and so many toys are dead ends as technology.

I want you to take a toy and break it and fix it. I want you to learn something from it, and I want you to think of what it can't do right now but could if you just had X, Y, or Z.

That's why I hope that Bespoke Robot Society will deliver robots that are not just toys. They should withstand toddler testing and enable real learning and extension.