The Smart Links for this week are divided into five big discussion topics, ranging chatbots to raycasting optimizations and level design discussion. Of course, you’re welcome to email me if you have any news or tips relating to game development and artificial intelligence!
Zabaware Wins 2007 Loebner Competition
Andrew over at grandtextauto.org points out that Robert Medeksza from Zabaware, creator of the Ultra Hal chatterbot assistant, has won the 2007 Loebner Prize, an annual “Turing Test” competition that’s been running for 17 years.
Try the online text-chat yourself to see if there has been any improvement in chatbot technology since Alice! Can you tell the difference?
A Level Design Analogy
Lewis over at the Teach Game Design blog muses about the role of level designers and how game AI fits into the picture:
“Fundamentally, level design is a limited, concentrated form of game design. The core mechanics of the game are already determined. The level designer is using them to create an episode that will have good gameplay, that will entertain in various ways. Gameplay always involves challenges and actions to meet those challenges, of course. This also involves goals, ways to achieve the goals, paths (such as corridors and rooms), appearances, and behavior of NPC’s and opposition (scripts to do better than the game AI can do on its own).”
A Level Design Analogy, teachgamedesign.blogspot.com
On Raycasting and Performance Optimizations
Pretty interesting discussion about the best data-structures to store polygons for efficient raycasting. This is useful for optimizing AI, as over 50% of your computation budget is spent on this typically…
Grid-Sort RayCasts by Charles Bloom
Perfect Spatial Hashing on Hugues Hoppe’s page
- Real-time Raytracing (PDF, 34.5 Mb)
Smoke and Mirrors: The True Art of Game AI
Microsoft Games studio Rare will be giving lectures in 2008 about game development, specifically this one:
“Game technology is constantly improving with the goal of creating the next must-have game, yet game AI technology has been pretty much static for a long time. How come this side of the games industry finds it so hard to move forward? There are numerous AI techniques that have been used in academic AI for years which could create the next big leap in game AI, so why aren’t they used? Should they be used?”
For anyone in the U.K., this is one event to look into!
LiveMove Goes Pro
The AiLive team sent in an email to announce a new version of their Wii gesture recognition product:
“LiveMove Pro is the new version of AiLive’s groundbreaking Artificial Intelligence product that allows games on the Wii to learn to recognize any motion. We have added full support for buttonless motion recognition, zero-lag play, and precise player synchronization.”
Stay tuned next Saturday for more smart links from around the web.