Hi,
I know you were probably looking for someone out of India or Pakistan to help you code this up for cheap, but I think they will fail to appreciate what makes this challenge awesome. I saw the CIG challenge over the past week, and I've been thinking about how I would approach the problem.
I think the hardest thing about this challenge is the localization, path planning and navigation. I am uniquely suited to solving this. My background is robotics, there are a number of sensing technologies that are used for localization, the least of which are cameras, due to the computational overhead. There are however solutions, the first of which is Visual SLAM. SLAM standing for Simultaneous Localisation and Mapping. Instead of a camera on a robot, we have the pixels from the game. This should be enough calculate optical flow, track features, recognize features. As our character travels through the game, a map is being built up and refined. We can then output this map, for future game play, removing that overhead. What this gives you in the end, a position of the character, a way to calculate the distance traveled through a map, a way to maneuver the character into certain positions etc. Which brings me to the next step.
This challenge also reminds of the Mario being played by a ANN generated by GA.
[login to view URL]
Which that, you have have access to the game data. The Visual SLAM, will provide a distance traveled metric like Mario.
I've ran out of space.