Why a puck for AugmentOS?

AugmentOS uses a compute puck.

All of your smart glasses apps run on a tiny compute puck, about the size of an airpods case, that goes in your pocket or your bag. This decision wasn't made lightly, let's explain why this is our approach.

Why not just run the apps on the glasses?

The best place to run smart glasses apps is on smart glasses themselves. Running all of the processing on the glasses themselves eliminates latency, power overhead, and complexity. However, we need smart glasses to be really light (<40 grams) and last all day (12+ hours). But to run all of the applications would require a powerful processor. This type of processor and its supporting components are heavy and large. It’s also power-hungry, and thus requires heavy, large batteries. People who try to make smart glasses that run the app on-board end up with glasses that weigh >100 grams, get hot, and die soon. So, for the foreseeable future, we won’t be able to run the applications on the glasses themselves.

So let's run the apps... somewhere else?

We need to offload the compute from the glasses to somewhere else. Where should it be offloaded to? That’s the critical question.

The Cloud

Why not just run all the apps in the cloud? We actually love this solution, but it’s not realistic today due to power constraints for data transmission and latency introduced by centralized cloud infrastructure. Consider the following wireless protocols and their power usage:

  • Bluetooth Low Energy – 10-100mW
  • WiFi – 200-500mW
  • 4G/LTE Cellular Data –  800-2000mW

Using 4G/LTE in smart glasses requires a massive amount of power – far too much to achieve lightweight glasses. 

WiFi is also too high power, and it’s often not available. Some might claim that new WiFi chips are getting lower power, and we could use a WiFi hotspot on our phones to connect our glasses to the cloud. This has two issues. First, it rapidly drains the battery of your phone and causes it to run hot – this is a non-starter. Further, Android and iOS don’t want the hotspot to run all of the time, so they constantly fight to shut it off. That means users are constantly turning the hotspot back on, which is a terrible experience. 

Bluetooth is potentially an option. One could write a phone app that connects to the glasses and passes data from the glasses to the cloud, and then passes data back from the cloud to the glasses. The issue here is latency. For today’s apps like live captions and translation, and tomorrow’s apps like augmented reality restaurant reviews, low latency is critical. Waiting for 100s of millisecond to send data to the cloud server and another 100s of milliseconds to get the data back is too slow to serve for many applications.

We’re optimistic that in the future, low latency, low power wireless protocols alongside internet upgrades like DePIN will help solve these problems. But AugmentOS is here today, and we’re taking a pragmatic approach, so we can’t allow rely on the cloud.

This is why AugmentOS doesn’t use only rely on cloud applications, but will soon support cloud applications for those use cases that don’t require edge compute and/or low latency.

Your Smartphone

The next option is your smartphone. It’s an edge computer, it’s powerful, it’s got wireless communication capabilities, and it’s already in your pocket – it seems like the perfect solution. We thought so too. But it turns out not to be the right solution.

We believe that smart glasses won’t be just a phone on our face – they’ll provide all kinds of new value. And that will be enabled by apps that are hands-free and/or proactive. In order to create an app ecosystem where the apps are proactive, we need multiple apps listening in to the sensor data – the context – from the smart glasses, all the same time. And we need multiple apps that are are able to push data to the smart glasses (to the display, speakers, etc.) at any time. The smart glasses of today use Bluetooth, and Android/iOS fundamentally don’t allow multiple apps to connect to one Bluetooth device at the same time. This is a major problem.

One solution is to create one app to rule them all – a central app that connects to your smart glasses, and then streams smart glasses sensor data to other applications, and receives data from other applications to show on the glasses. We built this out already in Android, it’s called the SmartGlassesManager. In Android, it can technically work, but it’s janky, unintuitive, and unreliable due to a myriad of operating system level restrictions. In iOS, restrictions on inter-app communications means it’s completely impossible to achieve. If we’re relying on a smartphone as the place that smart glasses apps will run, then we must support both Android and iOS, and it’s just not possible.

A further issue with using the phone is connection reliability. Android and iOS continuously kill background processes whenever they can. Despite years of effort and huge investment, no one is every able to make a wireless Bluetooth device that actually stays connected to your phone. Even billion dollar wearables companies make their devices more expensive and more power-hungry to add on-board memory, because they know their device will eventually disconnect from the phone and need to save data on-board until the user restarts the app and reconnects the device.

A Puck

Enter the puck. This is a small, lightweight device that lives in your pocket, purse, or backpack. It runs all of the smart glasses apps on a custom operating system – AugmentOS – which is designed for smart glasses. Multiple apps can run at the same time on your glasses, all with the guarantee that as long as the puck is alive, the apps will be running. It’s got extra battery and a 4G/LTE/5G module of its own so it doesn’t rely on your smartphone. 

It’s also more future proof. In order to create the fantastic AR + AI smart glasses future we’re all dreaming of, a lot of work is going to need to get done, and there will need to be some pretty drastic changes made to how the hardware works. We don’t know what all of those are going to be, but for example, we expect we’ll move from Bluetooth Low Energy to a much lower power protocol. Waiting for every phone manufacturer to implement the new protocol is going to take a long time. But if we’re using a puck, that change can happen quickly, enabling development to go from decades to months.

It’s also enabling for hardware companies. Everyone is going to need to offload their compute somewhere. If smart glasses hardware manufacturers can design their own puck for their own glasses, they can design their systems to create the best user experience possible – not hampered by whatever the smartphone industry has coalesced on.

In Conclusion

Overall, the community decided the best way to achieve a future where our smart glasses are awesome and help upgrade humanity for the better is by using a puck.

Want to help or learn more?