One of the few good things about lockdown and working from home has been having more time to spend with pets. But when the world returns to normal, people are going to go back to the office, and in some cases that means leaving dogs at home for a large part of the day, hopefully with someone coming into your house to let them out at the midday point.

What if it was possible for an A.I. device, like a next-generation Amazon Echo, to give your pooch a dog-training class while you were away? That’s the basis for a project carried out by researchers at Colorado State University. Initially spotted by Chris Stokel-Walker, author of YouTubers:How YouTube Shook Up TV and Created a New Generation of Stars, and reported by New Scientist, the work involves a prototype device that’s able to give out canine commands, check to see if they’re being obeyed, and then provide a treat as a reward when they are.

“We have developed an apparatus which uses machine learning to monitor and reward dogs’ positive behaviors,” Tom Cavey, one of the co-authors on the project, told Digital Trends. “If a dog displays a desired action for an ample amount of time, the device will release a treat. This is accomplished in real time by using a tiny, low-powered embedded computer called the Nvidia Jetson Nano. The device allows us to capture live video and use it in a machine learning model to figure out what the dog is doing in real time, totally independent of any local or remote connectivity. If the dog is displaying specific behaviors such as ‘sit’ or ‘lay down,’ a good behavior is detected, and a treat will be dropped from the device.”

A.I. is getting better at doing image classification all the time. But in this case, using a non-connected embedded device meant dealing with memory constraints and hardware limitations that hindered the researchers’ ability to use complex machine learning models. They utilized optimization and quantization techniques to reduce the model size, among other strategies, so that it could run on the Jetson Nano.

As to whether they have any plans to commercialize the technology, Jason Stock, co-author on the project, told Digital Trends: “Our approach consists of low-cost hardware and software components that would allow for commercialization of a relatively affordable product. We consider the possibility of developing this further given a greater budget for creating a better-functioning prototype, as well as improving the model and dataset to identify more behaviors — perhaps even those perceived to be negative.”

In other words, there’s a chance that this will wind up for sale eventually. But first you’ll have to sit, stay, and wait a while.

A paper describing the work was recently published on arXiv, the open-access repository of electronic preprints.