LogInOne of the team's developers posted an article on training AI agents to pick their own moves.
Tom Solberg from Embark Studios shared a new post on their team optimized animation workflow to save time on repetitive tasks. The developer noted that animation is a big bottleneck game development meaning that characters have to be designed and scripted manually which takes a lot of time if you don't have resources to grow your team.
Over the past two years, the team has been experimenting with physical animation based on reinforcement learning. "In short, that means we train physically-based machines to walk by giving them rewards for doing the right things - like virtual dog treats," wrote Solberg.
The team thinks that achieving good movement behaviors provides players with "more immersive and interesting gameplay, where the world becomes truly alive," with little or no pre-made animations.
Embark is using AI agents that observe their own bodies and the world around them, deciding how to proceed on its own. "This means that if the agent collides, is hit by, or generates some force by itself; it can adapt immediately for each unique situation."
Make sure to read the full post here. Don't forget to join our new Telegram channel, our Discord, follow us on Instagram and Twitter, where we are sharing breakdowns, the latest news, awesome artworks, and more.