Bipedal robots look cool, but are they the most efficient solution?
From the earliest days of science fiction, robots have often been depicted as bipedal and humanoid. That’s still true today, even though industrial robots have never followed this pattern. For true mass adoption of robots as consumer devices, however, major players such as Elon Musk are actively developing devices that stand on two legs. But is this really the best way forward for mass production consumer devices?
And, access all episodes of End of the Line on Engineering TV along with all of our other series.
* * *
Episode Transcript:
There are few technologies which have been more anticipated, predicted and desired than humanoid robotics. It’s been the stuff of science fiction since silent movies, and everyone from Fritz Lang to Steven Spielberg has presented a future where humans walk, talk and work with machines that look similar to us.
Boston Dynamics has famously capable bipedal robots, and Elon Musk is working on one as we speak. But from an engineering perspective, are we sure that walking upright on two legs is the smartest solution?
For humans, as for bipedal robots, stability requires active management. The autopilot on an airplane may seem to hold the aircraft on a steady course, but in fact it flies off course and constantly deviates from the desired trajectory, correcting constantly with servos adjusting control surfaces more or less all the time.
It is the same for humans. Our proprioceptors in our feet send unconscious signals to our brain that our center of mass is shifting outside our footprint, and we reflexively contract muscles as necessary to stay upright. We don’t have to think to stand up.
Bipedal robots don’t need a system quite as complex and can rely on things like gyroscopes and inertia sensors to keep the machine operating with its center of mass within its footprint area. It’s an incredibly complex problem, and Boston Dynamics publishes a humorous out-take reel showing the many times that development machines topple.
It’s remarkable that human beings can learn to do this so early, but given the high degree of hardware and software needed to achieve stability as it stands, why is it imperative to have robots that stand on two feet?
In the case of Boston Dynamics, their most commercially successful product is the small four-legged doglike machine, which can be fitted with a very dexterous arm and general gripper which can be used to open doors, among other things. Four legs are major advantage, offering intrinsic static stability and a lower center of gravity. And to do useful work with a gooseneck arm, keeping that weight low has beneficial effects on payload handling, as well.
I believe there is a general consensus that the Holy Grail of personal assistive consumer robots will require machines that roughly look and walk like a child. I think this is correct. I suspect that the first practical household robots will be like the Boston dynamics unit, a small dog with a goose-like neck and general-purpose gripper. With two optical sensors in the end effector for binoculars vision, and sufficient computing capability, this kind of machine ought to be able to do what we all wanted: do the laundry, clean the bathrooms and kitchen, take out the trash and generally tidy up.
Look at Star Wars: the eloquent, shiny and gold robot, C-3 PO, was bipedal. But R2D2, the useful robot, was shaped more like the kinds of units available commercially today. Two legs are good, but four legs may be better.
