So today it was back on to the mobile game and optimising some routing. It uses an a-star search with a modified Euclidean Heuristic, modified in a way that’s actually doing it wrong, but it does give reasonable results. I was doing length squared rather than length, which gives you quite wildly different answers when added together with the normal distance cost, even if you square that. So instead I decided to do it properly and change it to a Chebyshev. So the normal costs are just the distance either in a straight line or diagonal, then the heuristic is just a max of the absolute distance in x/z plus a couple of bits to handle the diagonals. It works quite well for some routes and not so well for others. Proof of the pudding will be in the iPhone4 testing. I’ll chat to some of my work colleagues who have a bit more experience on this.
Walked the dog. Went to the gym. Did body pump and a bit of running. Came back and carried on working, only just finished. I’m feeling very tired, time for wine, bath and bed. Oh my carpet shampooer and steamer have turned up. Will play with those the weekend.