they'll be quite broken in 3D
I was thinking of adding a special gene at the end that adds vertical motion normal or inverted to the .fw .bw the robot is doing based on where the robot is looking. The robot will also have to be aligned across the XZ axes. That means at times it will be traveling up and forward and at other times down and forward etc. I would need to also add some SS logic in the same gene to correct for .tieloc, .memloc etc. as the DNA will be "expended."
Anyway, the plan is to get rid of fixpos to 'nothing' Instead the robots will be able to fix to a shape.
Funny you should mention this Peter but I am trying to save your own bot. Your idea is to rotate the screen on startup so it is looking up across the z-axes. And rotate the eyes to point to the top of the head.
What I think will be cool is to have your Caterpillar robot that moves with ties to live on top of a shape "without water." That is, areas of the simulation where voluntary movement is restricted by the water level. This will tie into the new more realistic tide simulator. It will be the only known robot that lives on land.
Now last time I checked gravity is "Y" gravity, not "Z" gravity. Things tend to fall to the bottom. Or in the case of rotating the camera as you propose - to the front of the screen. Your very own robot will not work right if its eyes point to the top of the head. For eyes btw I am thinking cone shapes.
I did promise to keep DNA from DB2 working but I have seen other crawl simulators online and I think it will be cool to do in DB.