Designing a robot to mimic a species we're a little less intimate with could make for robots we actually care about.
Whether it's because they're taking our jobs or taking our lives, robots get their fair share of bad press. It's easy to forgot the promise of a robot in every home: They're supposed to make our lives easier, thanks to many of the same traits that make them effective human replacements. Who wouldn't want a Terminator to help clean the gutters and move furniture, if you could eliminate that whole time-traveling assassin part?
For once, this kangaroo robot by a company called Festo feels like something out of that shiny future. It's more of an engineering exercise than anything—and to be clear, mimicking the biomechanics and kinematics of a jumping kangaroo is quite impressive—but that might be why it's so cool.
This kangaroo bot exists just to exist, and its lack of an ulterior motive (as well as the fact that kangaroos are rad) makes me inclined it for a hop and, you know, just hang out. That's something that can't really be said for something like a Roomba, the sheer brilliance of the DJ version notwithstanding. Why?
The fact is, the robots that are commercially feasible and effective aren't the robots we thought we'd want. Everything from science fiction to kid's cartoons has focused on the idea of robots as things we interact with, things that we connect with and, hell, be friends with. It makes sense; as a whole, humans are fairly self-centered creatures who identify most strongly with themselves, so why wouldn't we want robots to be built in our own image?
The problem is the reality of development. As great as it'd be to have a robot friend play soccer with you, building such a beast is really hard. As NASA found in its own experiments with humanoid robots, a two-legged walking creature isn't the most effective solution for many tasks. And we've all seen that achingly brutal footage of poor, unstable Asimo falling down the stairs (which, admittedly, was years ago now).
The point is, the robots we might connect with, the C-3POs of the world, aren't what we're going to get. The bulk of the robots of the future will be designed for their task, not for us. Take the Roomba, which remains the most successful consumer robot around. Roomba, for better or worse, is designed like a whirring hockey puck because A) it works and B) it's the simplest design to bring to market. And because of that, even in a future where humanoid robots are cheap and abundant, it seems unlikely than anyone would decide to switch back to having a robot maid push a vacuum around.
Instead of a robot maid, you'll keep your Roomba and get a dishwasher that doesn't require you to scrub the dishes first. Instead of a robot chaffeur, you'll get an automated car. Your personal assistant bot is a screen on wheels, and your robotic surgeon looks nothing like the surgeon you know. We'll have self-assembling nanotrains and dragonfly drones, but are we ever going to get robots that feel like anything more than automatons? And no, I don't mean sex bots, I mean a robot that's neither faceless object nor creepy mimic.
That's what makes this kangaroo bot so interesting: It's built to be familiar. Of course, it's far from a frivolous design. As the Festo diagram above shows, a kangaroo's hop is a particularly complicated movement to reverse engineer. Combined with its ability to balance and reorient itself while still, as well as connect with a Myo armband, it's an impressive bit of work.
Kinematics and biomimicry remain huge areas of robotics research, and that's not likely to change; evolution is a particularly meticulous engineer, and one with ideas worth cribbing from. But compare this kangaroo to the products of Boston Dynamics, which is perhaps the best-known developer of bioinspired bots. Boston Dynamics' bots are stunningly capable, and yet they tend to hail from an uncanny valley carved out of a mountain of nightmares. This kangaroo though? I could chill with it.
Engineers everywhere would probably say something akin to "a tool for every job," and that if we want robots to cuddle with (like this baby seal), we should build them for that purpose. Both points are fair, and I think that in any case, we've all come to accept that the 1950s vision of a nuclear future isn't going to arrive with Rosie the robot maid.
Really, it's going to be a long time until a humanoid bot seems like anything but an awkward approximation of what a human is. But designing a robot to mimic a species we're a little less intimate with could make for robots we actually care about. Or, in the case of this kangaroo, a robot worth kicking it with.