I thought you guys might find this interesting. Support for OneShot ESCs was recently written for Tau Labs (http://taulabs.org) and it seems like most of the claims that it is good are largely subjective (which is perfectly fine). However, I wanted to really test the performance improvement with this different pulse scheme.
Here you can see the latency from an update from the gyro (MPU9250 IRQ) to the completion of the pulse is about 300 µs, which is much better than the several milliseconds with traditional asynchronous PWM.
I used our autotuning algorithm which estimates the system properties like the reaction speed of the ESCs on the same quad before and after engaging One Shot mode. The end result is a significant improvement in ESC latency as shown above. The results are statistically significant. Since the ESC latency is the prime performance bottleneck for quadcopters, this improvement will allow tighter PIDs and higher bandwidth control. That is the result of the autotuning algorithm - higher PID settings with OneShot than without.
For the longer writeup see here http://buildandcrash.blogspot.com/2015/01/oneshot125-quantitative-t...