Microsoft are now currently working on machines that will be fully autonomous; systems that actually learn from their surroundings and get smarter, without being controlled by a human.
How the system works
Microsoft’s autonomous systems platform uses an approach called ‘machine teaching’ which relies on programmers using a language called Inkling.
Using this language, programmers can show a system how to solve simpler problems and provide useful clues about what’s important. This increases learning speed and encourages the system to react in an educated and autonomous way.
Ashish Kapoor, Microsoft principal research manager gives us an example of the system in motion: “If I have the ability to spawn thousands of simulations at once and in each one, the pedestrian crossing the street is different and the curve of the road is different, suddenly the AI system is able to gather much more diverse experience in a short amount of time. Machine teaching helps the system learn in progressively more difficult steps.”
Getting the system ready for the real world
It wouldn’t make sense to let an intelligent control system make millions of mistakes in the real world while it’s still learning new algorithms.
So the Microsoft toolchain also includes AirSim, an open source simulation platform that was originally designed to teach drones, self-driving cars and robots in high fidelity simulated environments.
AirSim is a simulation technology that allows machines to learn in an environment that is highly realistic, but also safe.
Running these simulations in the Azure cloud means that the system can test thousands of different decision-making sequences in parallel to learn what does and doesn’t work, much faster.
It’s clear that Microsoft are at the forefront of these technological advances and we’re excited to see where they go from here.
If you’d like to find out more about Microsoft products and services, simply get in touch with the SCS Technology team on 01522 883636.