Original Post
There has been many many advancements in the field of game AI over the years and it seems that there isn't a game that doesn't have some type of intelligence in some form (NPC, physics, animation etc). I have been interested in Game AI after finishing the second module on Artificial Intelligence in my college. Although we used Prolog as the programming language, nevertheless, I now have a better understanding on the terminology used in the field. And finally the day I had to choose my dissertation topic came. I sat down and realized that the past couple of months (probably even years) I've been trying to gather a team to create a game only to be disappointed from the lack of interest of the party. So what better time to create a game and apply some of the techniques I learned than the final year project? But the techniques I learned are quite elementary level in comparison with todays standards. I've barely touch the surface of searching algorithms: Depth First, Breadth First, Iterative Deepening, Hill-Climbing, A* etc. We also did a little Knowledge Representation with Frames, Rule-Based etc. A little game playing: Minimax, Alphabeta etc. That's all. The past 2 weeks I've been reading non-stop about newer and better techniques. Every time I come across one that solves a problem a little better than the last it amazes me. Decision Trees? Influential maps? NavMesh? Group Formations? Height Maps? Hierarchical Pathfinding? Clearance-based pathfinding? And my personal favorite: Goal-Oriented Action Planning. The list never ends! I didn't have much time to come up with a game idea but luckily I didn't have to. All they needed in the first phase was just a title and a supervisor. So I simply submitted the most vogue topic I could have though of: "Multiagent Simulation". With this title I can come up with any idea of a game that uses multiple agents. And that is where this post comes in (finally). I've read similar questions in these forums but there was no answer that satisfied me. Not that I didn't note some of them. But what I really want is not what game to make. I want to know what is hot in Game AI techniques and try to implement this in a game. I don't care if the game has fancy graphics or anything. Just a basic proof of concept of the Game AI technique is enough for me. I recently came across another final year project that implemented GOAL and I was fascinated. I still remember a comment saying that if he could manage to create such a system then it wouldn't be difficult for him to wind up in the industry. I've been a programmer since my 16s, so I'm not worrying about the time it will take me to program it. After all, I could iteratively make it better. I'm mostly worrying in writing the report. Programming is the fun part anyway! So, what do you believe are the latest coolest techniques in Game AI? I would like to apologize for the length of this post. I wanted to introduce myself too as I will be around for a while. Regards, pek
