Amazon Astro Is One Small Step for Robotkind
Amazon's helper bot is a significant milestone for robots, but it's not the giant leap I was expecting.
Amazon's Astro is a landmark in the field of consumer robots. We're testing it right now as we prepare to give it a full review, but it doesn't actually have to be good to be important. Just the fact that it exists and is already in people's homes is a huge step that many ambitious robots have failed to reach.
I've enthusiastically covered the consumer robots on display at trade shows like CES for years. One of my biggest wishes for the industry is a fully capable bot that can act like Rosie the Robot from the classic cartoon The Jetsons -- one capable of cleaning up after me and doing basic chores.
Astro is cool to behold, but it's a far cry from Rosie. In fact, another Jetsons character inspired its name: Astro the robo-dog. Sure, Amazon's robot is the first of its kind with real potential to reach many customers. But -- for better or worse -- the pet-like approach seems more geared toward winning people's hearts (and money) than creating a highly capable helper-robot. Astro is a step forward in robot evolution, but I was hoping for a bigger one.
The ghost of Kuri
At CES 2017, I had the chance to play with Kuri, a robot from Mayfield Robotics. Kuri could autonomously map your home and travel to different rooms on command. Kuri had cute, expressive eyes. Kuri had built-in cameras that you could check from an app while you were out and about. Kuri would respond to voice commands and play games while you were around and act as a security sentry while you weren't.
Kuri was supposed to officially roll out to customers in December 2017 for $800. I saw it in action at CES in January 2017 and felt good about the company's chances to deliver on that promise. It was already rolling around well and responding to commands. The app was up and running. Mayfield Robotics was a startup, but it was backed by appliance giant Bosch.
Kuri and Astro might have been friends.
Instead, in July 2018 Mayfield Robotics officially canceled production of Kuri. Now, more than five years after I first saw Kuri in action, we have Astro. Amazon first showed off Astro at a fall 2021 press event. It's now rolling out to customers on an invite-only basis for $1,000. I've seen Astro in action -- not under the tutelage of its design team at a press event, but navigating the CNET Smart Home on its own.
In terms of features, Astro has expressive eyes and responds to voice commands. It has built-in cameras and can act as a sentry while you're away. It maps your home and can travel to different rooms. In a lot of ways, it is the resurrected ghost of Kuri, just made by a tech giant instead of an ambitious startup.
Kuri could have been a win for a little guy in a way that Amazon's Astro never will be. Some of Astro's features are beholden to a subscription to one of its parent company's subsidiaries: You need a Ring Protect Plan subscription to unlock Astro's abilities to patrol your home on its own.
Astro's connection to Ring and Amazon also raises legitimate privacy concerns. Ring's messy history with privacy policies and police partnerships is well documented at this point, and Astro connects the company to a roving bot with multiple cameras that also knows your floor plan.
Astro can patrol your home while you're gone.
Amazon, to its credit, has put some safeguards into place. Astro can recognize faces, for instance, but it processes that info locally. Likewise, it stores most of its mapping information on-device. But if you want to check your camera while you're on the road, that necessitates using the cloud.
That's not to suggest Kuri wouldn't have used the cloud; it almost certainly would have. In fact, it's highly possible that a startup like Mayfield Robotics would have had less capacity for high-level security than a giant with near-limitless resources like Amazon -- but it also would have been less of a target.
Plus, I'd have been less concerned with how Mayfield Robotics was using my data. Again, Amazon is saying the right things on this front for now, but Astro requires you to trust a giant company with highly private data.
The Astro of the future
Astro isn't actually the only available consumer gadget that makes use of robotics. Sony's Aibo is a cute robot dog. Jibo was a robot that could answer your questions, but couldn't move. Ubtech has a yoga-teaching robot called the Lynx. Robot vacuums similarly have a lot of built-in navigation tech and higher end models can map your floors, too.
Compared to the rest of the field, though, Astro is the first major consumer product that truly resembles the traditional depiction of a robot. Jibo was mostly a dressed-up smart speaker and didn't last long on the market. Lynx and Aibo are more or less toys. And robot vacuums are too limited in purpose for me to think of them as actual robots instead of utilitarian gadgets with robotic navigation tech.
Sony's Aibo is certainly cute.
Given its lead on the available field, you might think my hopes for a more ambitious Astro stem entirely from science fiction and classic cartoons (and Astro self-consciously echoes those depictions). But it's really for a more mundane reason: I want robots to be genuinely useful. Attaching a vacuum to Astro's base, for instance, would be a simple add that could greatly improve its functional versatility.
As it stands, Astro can deliver drinks and throw a dance party to entertain the family -- but once you get over the gimmicks, its most useful feature is as a security bot. If it had a vacuum on the base, it could also legitimately help out with the chores. The upcoming Roborock S7 MaxV Ultra will attach a camera and speakers to a robot vacuum, which suggests the basic combination of vacuum and camera can work.
Beyond that, for Astro to be a truly capable robot helper it would need more robust object recognition and a way to grip and pick up everyday items. Imagine if Astro could empty the dishwasher or tidy stray toys or laundry. Astro can bring you a beer... but only if someone puts on its cup holder attachment and sends it to you with a command. I want a bot that can do all of those steps.
While this would certainly be a stretch in terms of the technology that has come to market, Samsung showed off a concept last year called Bot Handy that could do just that. At CES in 2018, I saw an exciting robot called Aeolus that could recognize and pick up objects. Aeolus was almost as big as a human, and in theory it could even grab and use a vacuum cleaner.
The Bot Handy could do its share of the chores.
Bot Handy could pour the wine and fold the laundry, and Aeolus had all that and more. Even at the time, though, Aeolus was being strictly controlled by a programmer using a computer, so it's unclear how close the tech was to actually working in a home environment. We haven't heard from Aeolus since. As for Bot Handy, Samsung is a big name, but the company has a history of showing off cool robot concepts that, for the most part, stay concepts.
We've also seen a robot named Sophia hold a conversation and others from Boston Dynamics pull off all kinds of acrobatic feats. The pieces for a walking, talking helper bot currently exist in some experimental stage or another.
The Misty II could provide the test bed necessary for designers to fine tune and bring some of these ambitious ideas to life. Misty II is aimed at developers and engineers instead of consumers. It already has some navigation abilities built in, but its pieces are designed to be swapped out and its software is designed to be reprogrammed. The idea is that Misty can provide a sort of blank slate for folks looking to test cool new robotic tech on a functional model.
Astro's disappointing present
Given the ambitious tech I've seen on display over the years from these concepts, I'm slightly disappointed in Astro as a technological leap, because we saw most of its capabilities in action five years ago. Again, Astro is in homes -- something few developers have been able to accomplish. That said, Astro is invite-only, and Amazon hasn't settled on concrete plans for a full commercial release. Many Amazon products don't make it past this phase, such as its glasses and rings with its assistant Alexa built-in.
Amazon's experimental branch does have its share of success stories, too -- most notably the Amazon Echo. That product went on to redefine the smart home industry. At the time, it was also a safe step past ambitious products from startups that had tried and failed to do the same thing.
I wanted Astro to be more ambitious, like Misty.
At the moment, Astro feels experimental in action. It struggled to map the floors of the CNET Smart Home until we blocked off our railings and closed the blinds so the hardwood floors weren't as shiny. That said, the original Amazon Echo wasn't nearly as polished at responding to voice commands as today's Alexa-enabled smart speakers.
If Astro receives a full release, it'll cost a whopping $1,500. It could end up being an overly expensive flop that doesn't make it much further than Kuri, given the invite-only nature of its current release. Worst case, a failure from a giant like Amazon could discourage smaller manufacturers from even attempting further headway in the field.
The best case scenario for Astro is exemplified by the Amazon Echo as a case study. It's currently neat to see Astro in action as it can already navigate somewhat fluidly (now that we've helped it with the initial mapping process). If it's a success, Amazon could proceed to polish off the rough edges and Astro could help lead the way into ambitious new territory.