An avid programmer since the mid-80s, I have used and abused Prolog, LISP, Smalltalk, Java, Erlang and Elixir. Years spent in corporate IT departments and with startups have taught me that strong domain models are the beating hearts of successful software projects.
I am currently software architect at Starlit Software. I live and work in Portland, Maine.
In my spare time, I train obsessively at Aikido of Maine, and organize the local Erlang/Elixir and Elm meetups.
Could your Lego robots learn from experience? How would you go about it? I was asked these questions at last year's conference and I did NOT have good answers. I mulled over them for weeks and felt hopelessly stuck. But, as luck would have it, I ran into the "predictive brain" model of cognition. It provides strikingly beautiful answers to the learning problem.
I had found a deeper model! So I set out to refactor my robots's cognitive architecture and incorporate prediction and learning (from prediction errors) as core domain concepts. This presentation will demo the outcomes, retrace my steps, and show how they are illuminated by Domain-Driven Design.