I have posted an NXT 1.0 version of the NXT Segway with Rider project (see previous blog entry on the NXT 2.0 version).
This version uses the standard light sensor instead of the color sensor, the taller balloon tires, and the NXT-G 1.x software (tuned for the 1.05 firmware). Although the control program runs significantly slower than the 2.0 version (the newer 2.0 software and firmware has reduced the overhead of the sensor and motor blocks considerably), the 1.x version fortunately is able to make up for it by using the light sensor's "Raw" output (not available on the 2.0 color sensor), which has much higher resolution, which makes it easier for the robot to maintain an accurate on-center position, despite the slower sampling rate.
So the best NXT-G configuration for this sort of task (or similar, think line following...) would actually be the NXT 2.0 software and 1.28 firmware but using the old light sensor in Raw mode, so you would get both the high speed and the high resolution. If anyone tries this starting from my programs, you will need to re-tune the PID gains to account for the higher sampling rate, since the calculations do not take into account the time slice...
This version uses the standard light sensor instead of the color sensor, the taller balloon tires, and the NXT-G 1.x software (tuned for the 1.05 firmware). Although the control program runs significantly slower than the 2.0 version (the newer 2.0 software and firmware has reduced the overhead of the sensor and motor blocks considerably), the 1.x version fortunately is able to make up for it by using the light sensor's "Raw" output (not available on the 2.0 color sensor), which has much higher resolution, which makes it easier for the robot to maintain an accurate on-center position, despite the slower sampling rate.
So the best NXT-G configuration for this sort of task (or similar, think line following...) would actually be the NXT 2.0 software and 1.28 firmware but using the old light sensor in Raw mode, so you would get both the high speed and the high resolution. If anyone tries this starting from my programs, you will need to re-tune the PID gains to account for the higher sampling rate, since the calculations do not take into account the time slice...