The creepiness is pretty fascinating… There’s been endless studies and research into human/robot interaction, and it influences design a ton. There’s a reason humanoid robots usually move pretty much like humans- it’s not optimal, but it’s more familiar. Having them definitely have a front and back and similar movement restrictions makes them more intuitively predictable for people working around them, you can predict more easily where it’s likely to move next, what it’s doing. And it just makes it more comfortable to deal with them, removes a ton of the scare factor. But it actually holds back the design and capability. Bipedal robots in human/robot shared spaces are always likely to be less overall good than we could make them in a robot-only or robot-and-robot-expert space, because they’ll compromise so they don’t scare your gran or so that you don’t constantly get in its way.
This one’s thrown a load of that out, and the result is really unsettling. But also harder to interact with- which way is it facing? Both ways. What way will it go next? Who knows. Especially since they’ve incorporated the “change direction” with it actually moving so it’s not even like it reverses the legs then moves, it steps “backwards” then after a step it’s going “fowards”. Makes for a much better robot at the expense of making it weirder but also creates problems like- well, it shouldn’t bump into you or step on you because it’ll be full of sensors, but if you get in its way then it has to stop.
But that’s partly because when they were rolling out asimo and stuff like that, the whole thing was weird so they were fighting that and trying to make them comprehensible and acceptable. Making them a bit less competent helped with that too, that’s a whole fascinating field of study, humans really like clumsy robots, noisy robots, robots that look a bit DIY. We want them to have flashing lights and go biddibiddibiddi. (They had an interactivity testbed at my old work where they went through and replaced a load of stepper motors and controllers with newer, quieter ones, with no other changes, expecting positive results as it meant it wasn’t making constant horrible screeches and robot wasp noises- but the less motor noise it made the weirder and faker it seemed, the interactivity survey scores got worse)
But now as we get more used to it, you can be weirder without such extreme reactions. It still ABSOLUTELY had to have a head or a face though, turns out humans just don’t know what to do with a robot if it doesn’t have a head or a face, it massively undermines the whole thinking process. You’re more likely to be surprised when they move, more likely to have panic or fight-or-flight reactions. But at hte same time, uncanny valley means it’s better to have, say, a screen with a smiley face than it is to have hte really interactive, emotive human-analog faces that they’ve tried. We’re used to interacting with screens, after all. And having it on a head scores far better than having it just on the torso, that causes a lot of bad response.
This’ll keep changing the more used we get to it, biped robot design’s basically as flexible as the human brain. It’s just… super nerdy interesting imo. Especially since the major reason for this sort of flexible bipedal robot is so that it can take people’s jobs without rebuilding factories and warehouses and such 😉
