|
From: David S. <dsa...@te...> - 2014-02-10 20:02:17
|
Hi Bernhard, I will check out the r1-3-1 branch (testing will take several days). Also I'll look into the quaternion for car body rotation that you mentioned, because it looks interesting. The patches show my attempts to make TORCS better; the good patches get used, the others are for conversation :) The following are my qualitative observations of random TORCS stuff: I was looking into the moment of inertia, still am, and here is my latest comment: "max steer speed" feels like it augments the moment of inertia. My favorite value is 12 deg/s; this does not make the steering slow, it seems to filter out "twitchy" steering. Press 4 on the keyboard while driving and look at the steering forces vector. car4-trb1 <section name="Steer"> <attnum name="max steer speed" unit="deg/s" min="1" max="360" val="360"/> In torcs/src/drivers/human, after some calculations: car->_steerCmd = leftSteer + rightSteer; Is it easy to make a purely linear steering formula that puts linear mouse movements into linear steering (rack and pinion)? I need some hints to get me on the right track. I read somewhere that TORCS determines the engine speed from the wheel speed from the car speed. I've noticed that the engine loses rpm while sliding sideways as a function of car angle and speed. Maybe the car would drift better without the function of car angle affecting RPM. Sincerely, David Savinkoff |