Hi, Track space transform was used in Downforce. AI Game Programming Wisdom 2 describes it, but not in detail. Basically, the track consists of a number of track lines, more or less evenly spaced. We can transform the 2D position of the car, into width position and length position to represent the position of the car, between two track lines. width and length pos are supposed to be 0..1. So, what we have to work with is 4 2D points, two for each track line, and the position of the car that is between the two track lines. There's no guarantee that the 4 edges formed by the 4 points are perfectly parallel or perpendicular. The concept is described here: www.fdaw.unimaas.nl/education/4.5GAI/slides/Racing%20Vehicle%20Control%20Using%20Insect%20Intelligence.ppt I would like to generate width pos and height pos with the least operations because they'll be fed at each step to a neural network. What methods would you recommend using? Thank you Alex