For every tech success story, there are countless projects that slam headlong into the brick wall of reality. Apple’s electric vehicle ambitions are one of the most recent — and, frankly, best — examples of a project failing in spite of seeming to have everything going for it.
The jury is still out on the ultimate fate of the Vision Pro, but at the very least, Apple’s mixed reality headset demonstrates that the company isn’t afraid to keep trying where pretty much everyone else has failed. With the Apple Car firmly in the rearview, the company is reportedly exploring yet another notoriously difficult path: home robots.
The category is both unique and uniquely difficult for a number of reasons. One thing that sets it apart from other categories is the fact that there’s been precisely one success story: the robot vacuum. It’s been 22 years since the first Roomba was introduced, and for the past two decades, an entire industry (including iRobot itself) has been chasing that success.
iRobot’s inability to strike gold a second time is not for lack of trying. In the nearly quarter-century since it introduced Roomba, it’s given us gutter clearers, pool cleaners, lawn mowers and even a Roomba specifically designed to remove screws and other hardware detritus off garage floors. In spite of those efforts, however, the company has fared best when it focused its resources back into its robot vacuum.
The robot vacuum succeeded for the same reason any robot has ever succeeded: It was a product built to perform a single in-demand task repetitively to the best of its ability. To this day, vacuums are the battlefield on which the home robot wars are fought. Take the well-funded Bay Area startup Matic. The former Google/Nest engineers who founded the company believe the next breakthrough in the home will be built on the foundation of robot vacuums. Their case, in part, is that iRobot effectively painted itself into a corner with its puck-like form factor.
Those early Roombas weren’t built with today’s sensing and mapping capabilities in mind. Matic believes that by simply making the robot taller, you dramatically improve its vantage point. This was also the driver behind the most interesting innovation found on Amazon’s Astro home robot: the periscope camera.
The fact is that home robot functionality is severely hampered by form factor. The hockey puck design that’s prevalent across robot vacuums isn’t ideal for anything beyond the core functionality it’s built for. To effectively perform more of the sorts of tasks people might desire in a home robot, the hardware needs to get more complex. Mobile manipulators are a great moving target. That is to say, if you want a helping hand, a hand is a good place to start.
Like so many other things in this world, however, mobile manipulators are deceptively difficult. In fact, industrial robotics haven’t cracked it yet. Big, bolted-down arms are common in manufacturing, and wheeled autonomous mobile robots (AMRs) like Locus and Kiva are common in warehouses, but the middle ground between the two hasn’t been firmly established. This is a big part of the reason the human element remains important in that world. It’s a problem that will be solved soon enough, but it seems likely it will happen with these more expensive industrial machines well before it makes its way into more affordable home robots (as a rule, corporations generally have deeper pockets than people).
This is also a big part of the reason many are championing the humanoid form factor in the workplace (human beings, after all, offer a kind of mobile manipulation). But that’s a longwinded think piece for another day.
Mobile manipulation isn’t entirely out of reach for home robots. Hello Robot’s Stretch is probably the most compelling example at the moment. Rather than a humanoid form factor, the robot looks like a Roomba with a pole mounted in its center. This houses both an imaging system and an arm that moves up and down to clasp objects (dishes, laundry) at different heights. Of course, some tasks are more easily accomplished with two arms — and suddenly you start to see why so many robotics firms have effectively backward-engineered humanoids.
In its current form, Stretch is prohibitively expensive at $24,950. That’s likely a big part of the reason the company is selling it as a development platform. Interestingly, Matic sees its own robot as a kind of development platform — using vacuuming as a gateway into additional home chores.
Another issue with Stretch is that it’s teleoperated (the company sent us a note after this published stating that some developers have created autonomous functions). There’s nothing wrong with teleop in many scenarios, but it seems unlikely that people are going to flock to a home robot that’s being controlled by a human somewhere far away.
Navigation is another key barrier to the home. Compared to warehouses and factories, homes are relatively unstructured environments. They differ greatly from one to another, lighting tends to be all over the place and humans are constantly moving stuff around and dropping things on the floor.
The world of self-driving has faced its own obstacles on this front. But the key difference between an autonomous robot on the highway and another in the home is that the worst the latter is probably going to do is knock something off a shelf. That’s bad, but very rarely does it result in death. With self-driving cars, on the other hand, any accident represents a significant step back for the industry. The technology is — perhaps understandably — being held to a higher standard than its human counterpart.
While adoption of self-driving technologies is well behind the curve that many anticipated, largely for the above safety reason, many of the technologies developed for the category have helped quietly kickstart their own robotics revolution, as autonomous vehicles take over farms and sidewalks.
This is likely a big part of the reason it might view home robots as “the next big thing” (to quote Bloomberg quoting its sources). Apple has no doubt pumped a tremendous amount of resources into driving technologies. If those could be repurposed for a different project, maybe it won’t all be for naught.
While the reports note that Apple “hasn’t committed” to either the robotic smart screen or mobile robot that are said to exist somewhere inside the company’s skunkworks, it has already put Apple Home execs Matt Costello and Brian Lynch on the hardware side of things, while SVP of Machine Learning and AI Strategy John Giannandrea is said to be involved on the AI side of things.
Given the proximity to its home efforts, one can imagine the company working on its own version of Amazon’s Astro — though that project currently exists as more of a cautionary tale for the time being. The project has been hamstrung by high cost and a lack of useful features to justify it. The system also effectively served as a mobile Alexa portal, and home assistants have largely fallen out of fashion of late.
Apple does have some robotics expertise — though nothing approaching what Amazon has on its industrial side. The company has been involved in the production of robot arms like Daisy, which salvages key metals from discarded iPhones. That’s still a pretty large leap to a home robot.
Perhaps the company could take a more Vision Pro-like approach to the category, which has a heavy focus on developer contributions. Doing so, however, would require an extremely versatile hardware platform, which would almost certainly be cost-prohibitive for most consumers, making the Vision Pro’s $3,500 price tag look like small potatoes.
Comment