An Apple Robot is less likely to happen than an Apple Car

With Apple now having dragged its decade-long car project to the trash can, the company is reportedly now working on another skunk-works project: an Apple Robot.

Both self-driving cars and home robots have a lot in common. They have long been a science-fiction dream, and nobody has come anywhere close to realizing that dream. But a truly useful and capable domestic robot is an even bigger challenge than an autonomous car …

The self-driving car dream

The self-driving car dream dates back to at least 1935. Some 89 years later, we’re not there yet.

That dream is getting into a car which has no human driving controls, telling it where we want to go, and then sitting back and relaxing until we reach our destination. No constraints on where we can go, the types of roads which can be driven, the weather, the traffic … just get in and go. That’s known as Level 5 self-driving.

But despite more than a decade of trying to get there, and many years of using the Full Self-Driving (FSD) label, Tesla is still only at Level 2 – essentially combining a GPS route with automated lane-keeping and adaptive speed control, with the human driver needing to be ready to take over at any moment. Sure, the latest version can cope with a lot of situations along the way, but it’s the requirement for the driver to be ready to take over at a moment’s notice which means it’s still L2.

Mercedes has hit Level 3: totally automated driving without any human intervention, but in very limited circumstances, and with a human driver ready to take over with a little notice. In this case, the ‘limited circumstances’ are autobahns (freeways) in slow-moving traffic up to 60kph (37mph). The human driver notice period is 10 seconds.

Google’s Waymo has reached Level 4 – fully automated driving from A to B with zero human intervention within certain constraints. In Waymo’s case, those constraints are geographic: the cars can only operate within a small area which has been electronically mapped in advance with millimeter precision.

It was recently estimated that Level 5 won’t be achieved before 2035, by which time Apple would have been working on the project for more than 20 years. And nothing short of L5 would have made sense: the company would otherwise have been entering a crowded market not known for high margins. Getting out was likely a smart move.

The domestic robot dream

The domestic robot dream dates back even further, to at least 1920. Over a century later, we’ve made even less progress with this than we have with Level 5 cars.

The dream is effectively replicating the capabilities of a human servant from the Victorian age. A robot which can take care of a wide variety of chores around the home, from cleaning the house to fetching us a drink.

One of the most advanced attempts so far is Figure 01, a robot whose intelligence is powered by ChatGPT. Here’s a look at it in action:

As a demo, it’s impressive. The robot recognised some objects, and intelligently figured out what to do with them.

But the environment it was faced with was incredibly simple. There was a single item of food, and that was a very colorful Apple in an otherwise monochrome environment. There was one plate and one mug that needed to be moved, and only one possible place to move them. From the way it just dropped them when they were in vaguely the right place, I also don’t rate the survival chances of crockery and glassware for more than a few days!

This isn’t even the early Tesla of robots. This is at best adaptive cruise control being used on a one-lane road with no junctions.

An Apple Robot isn’t happening anytime soon

Apple is not a company that dives in early with cutting-edge tech. It lets other companies race to be first to market, while it works on the best possible implementation.

I mean, take something as simple as a folding smartphone. Samsung launched the first of the Galaxy Fold series back in 2019, and we’re not expecting the first folding iPhone until 2027 at the latest. We’re going to see a Samsung Robot way before we see an Apple one.

Then we have to consider price. Sure, Apple sells premium products, but it’s still essentially a mass-market consumer electronics company. It sells products which are priced somewhat higher than competing ones, but which are still within reach of ordinary people. It’s not selling niche products for the super-wealthy. The Apple Car could have been a BMW or a Mercedes, maybe a Porsche – but it couldn’t have been a Ferrari or a Lamborghini.

Think about Vision Pro, for example. Everyone knew it would be advanced. Everyone knew it would be expensive. But there were still audible gasps when the company revealed the starting price of $3,499. The consensus view is that this largely a tool for developers to play with the tech, and start creating apps for more wallet-friendly models to follow.

So even if we do see an early and super-expensive Apple Robot Pro launch first, that too would be merely a precursor to a true mass-market product.

What boxes would an Apple Robot need to tick?

I don’t think Apple’s going to make a toy. There are going to be a ton of so-called home robots which can carry out very limited tasks in very controlled conditions, and as much as we might enjoy them for the gadget factor, they are going to be like today’s VR headsets: Most of us will be wowed by them for a short time before the novelty wears off.

So, let’s forget solutions in search of a problem, and a possible Vision Pro equivalent, and look ahead to the boxes which would need to be ticked by a true consumer version of an Apple Robot.

Functionality

It has to be genuinely useful. Washing dishes by hand may make for good demo, but most homes have perfectly good dishwashers. Hi-tech homes also already have robot vacuum cleaners, as well as Siri to do things like switch lights on and off. So it can’t just do one or two tasks, and it can’t just to things we can do with existing tech.

I’d say a worthwhile functionality set for a true domestic robot would be something like this:

• After a meal, put the crockery, glassware, and cutlery into the dishwasher
• When the dishwasher finishes, put everything away where it belongs
• Pick stuff up from the floor (like kid’s toys) and put them where they belong
• Tidy up generally (eg. fold blankets, put cushions back on couches, etc)
• Collect laundry, and put it in the washing machine
• Transfer it to a dryer, or hang it up to dry, when done
• Make beds
• Feed pets
• Water plants

Things like cooking simple meals, cleaning windows, and taking out the trash would be much further down the line, I think.

Usability

Setting up has to be something mass-market consumers do, without anything approaching writing Shortcuts. The most usable UI I can think of for a domestic robot is to show it how to do something. It watches while a human carries out each of the tasks, and then replicates them.

There will probably be some unavoidably tedious aspects to this, like identifying each type of plate, glass, and so on, and showing the robot where each needs to go, but that would be fine as a one-off task.

Cost

This is a tricky one, as there’s currently no basis for comparison.

One potential comparator would be the cost of a human cleaner. How much would it cost per year to pay a human to carry out the same tasks? And what would be the payback time for an Apple Robot?

If we viewed the actual value delivered from say 1-2 hours a day, and a payback time of several years, then we’d be in five figures. But of course even the ~10% of US households who have a regular cleaner couldn’t run to paying that much. So I’d say that we’re really looking to hit $9,999 as a maximum.

Even if it were technically possible to build this today, we’d be adding a zero to that, and still adding in some multiplier.

You’d need something more advanced than the Vision Pro camera set just for the eyes alone, and that’s before we get into all the proximity sensors, pressure sensors, and so on needed for the hands. We’re probably at $5k plus before we even add the first motor to enable it to actually do something.

This is well over a decade away – if it ever happens

So this type of capability, at a price which comes anywhere near even Apple consumer pricing, is almost certainly well over a decade away. Probably two.

We’ll get a Level 5 self-driving car before we get an Apple Robot.

So does it really make sense to give up on one project, which might have been realizable in ten years, to instead work on another, which almost certainly isn’t?

I fully believe Bloomberg that Apple is playing with this idea, but my bet is that, like the car, it never materializes.

What do you think? Please take our poll, and share your thoughts in the comments.

Image: Figure.ai