When working with the test framework of the AI Sandbox, it's often convenient to rename a unit or functional test temporarily while working on some feature so it shows up on its own at the top in the test runner. Some locomotion tests in particular, show up as AAA_Locomotion while we're fixing or improving things! This serves as a constant reminder of our ultimate goal of figuring out how to make believable characters with high-quality animation yet responsive behaviors.
At the GDC last year, I gave an intense 30 minute overview (condensed from a 1h30 masterclass) about the various ways you can implement character locomotion in your game, and what's next if you'd like to improve your system. The part about the future of locomotion for the games industry certainly created a bit of a stir (see my report at the time, Motion Planning for Fun and Profit!) and raised some eyebrows too: “Do we really need better locomotion in games?” At the other end of the scale, it's been fascinating to hear reports and questions from developers who are already convinced and embarked along in this journey too!
Since this is one of my favorite topics and I've received multiple questions over the past months that I can't do justice to by email, I'm going to dig into the details in this article. The rest of this review I'll dig into what we've learned on the subject since my presentation at the GDC. At the end of the article, I'll also talk about what we're aiming for with the AI Sandbox in the near future based on that. Multiple programmers on the AI Sandbox have worked part-time on this system (approx. 6 months of work total), and I'd like to thank them for their contributions and insights: Jad Nohra, Kenneth Shor, Piotr Trochim, and Kyle Walsh.