Hi all, I worked recently on bikes AI and found a lot of "specific" problems in this field compared to generic racing AI(cars) programming. Is anyone aware of some paper/article about this? I'm in search of new and fresh ideas ... ;) Thanks in advance, Paolo

You need to make the bikes move in curves. Therefore, any AI for controlling motorbikes would also need to be able to handle curves. Write some routines to figure out which point on the track to head for next, then get the AI to calculate, plus some errors, a number of routes to get there. Each route is a series of curves. Pick the best route and take it.

Well,
I found some different challenges in more than one aspect. One for sure is the steer managment that for a motorbike game should be more smooth and constant. With a car you can accept sudden correction while with a bike you need to keep the lean more constant than possible. This could be quite easy to handle on curves with constant radius, while is more difficult with a more "generic" spline (in which the radius can change continuously during the curve).
I found some empiric solutions to these problems and solved them with a lot of tuning. I just wonder if someone else found my same difficulties and I'm curious to see their own solutions. Or if someone is aware of some paper (even just theoric) that could fit to this kind of problems ... .
Of course all is dependent of the physics model the game is using. I guess that for an arcadey game things between cars and bikes could be more similar ...

Paolo