If anyone wants to try this, there's now a tar file here, with the edit mentioned below.
http://www.cs.bham.ac.uk/research/projects/poplog/packages/popracer.tgz 1.4MB To run it you first need to relink poplog to include motif, using: $popsrc/newpop -link -x=-xm I've just tried popracer again. I edited popracer/main.p file to include uses rclib, rc_window_object, rc_buttons; then ved main.p Compile the file ENTER l1 Produces in a new ved window: PopRacer v1.08 (c) Team Popcorn 2005 - Loading Modules...Final Release by Michael Brook * Neural Network and Genetic Algorithm Engine OK * Physics Engine OK. - Using Gravitional Constant:9.81 - Using Friction Coefficient:0.002 * Bezier Engine OK. - Track Editor OK. Welcome to PopRacer by Team Popcorn 2005 *** Type help to list commands *** Pop Racer Command Line: Command? Then, following the instructions in userguide.pdf Command? help Produces a long list of commands, including: ? run Runs the simulation ? setcars <num of cars> Choose the number of cars to be simulated ? loadcars <filename> Load a previously-saved population ? settrack <trackName> Choose a previously-saved track .. ? createtrack Launch the track editor to create a new track ? savetrack <file name> Saves the track in the track editor to the file specified .. All very promising! I typed ? run That brought up a large window showing a figure 8 racetrack, a small control panel with run, stop, quit, buttons and three boolean toggle switches Animate: F Paths: F Graphs: F Clicking on start produced ***** Started SimEngine ***** ** Waiting for User. * Creating New Brains... *** Running Simulation... *** * Generation 1 Start... But no visible action. So I toggled the three Fs to Ts and clicked on Start again. That drew a small picture on the racetrack, which I dimly recognized as a picture of a racing car viewed from the top. (Using low resolution pop11 turtle graphics I suspect?) Thereafter repeated (fast) clicking on "start" produced many small changes as the car moved leaving a messy picture (presumably the original redrawn with a small change). After clicking about 50 times I began to see signs of trails left by moving objects. After many more clicks they began to separate out showing numbers. e.g. c17, c30, c46, c47, c48... and a few of the cars started moving non-randomly, but in different directions. I assume something should have started training for more than 1 cycle at a time. Anyhow it looks as if the code basically works but needs some changes possibly because of changes in pop11 since 2006. I'll leave it to Mark, or anyone else who wishes to read the code and experiment. I may try again after catching with a large backlog. Aaron
