Gradient Accuracy Test
Test procedure
This test measures the accuracy of the 20Hz GPS only gradient output of the SPEEDBOX. The test simply comprised of fitting a standard production SPEEDBOX to a road car, and driving at a relatively constant 70kph up and down the same section of highway. The highway selected was public, and lined with trees, but free from significant buildings and with a good view of the sky. The same line was driven by trying to keep the passenger side road wheels aligned with the road markings. The antenna was fitted to the center of the roof, with the SPEEDBOX in the passenger foot well. All the serial SPEEDBOX data was logged using a VIDEO4 unit. The VIDEO4 was used so the video footage could be used to check that the correct line was taken on each run:
The section of road was approximately 1km long, with convenient turning areas each end. The road was driven up and down 10 times:
The measured gradients in each case are shown below:
The correct gradient is taken as the average of the readings, and then in each case we calculate the error, this is shown in the graph below:
To make the distribution of error clearer, it is shown in the histogram:
The above results are generated with the gradient measured at 20Hz and filtered over a period of 2 seconds to reduce the sensitivity of minor changes in line. However the accuracy of the results can be significantly improved if a slower response is acceptable. If we increase the filtering to 20 seconds then we get the following result:
So in this case the error has improved by a factor of 3, giving a standard deviation of error of just 0.006 degrees