Statistical evaluation of data from tractor guidance systems

Statistical evaluation of data from tractor guidance systems

Statistical tools are discussed for the analysis of data collected from tractor guidance systems. The importance of both accuracy and precision is discussed, and statistical
tools for analysis are considered which incorporate important features of the data.
In particular, accuracy is modelled using a generalized least squares model incorporating autocorrelation, and variances (inverse of precision) using a gamma generalized linear model. The methods are applied to data collected during an experiment conducted with a
Trimble receiver used with a Beeline tractor guidance system. Three different scenarios are
considered, then compared: a tractor simulating ploughing a field; the tractor pulling a
plough with the receivers on the tractor; the tractor pulling a plough with the Trimble
receiver on the plough. The change in the precision and accuracy between the scenarios is
discussed. Data were recorded over repeated swaths for each scenario. After discussing
specific statistical techniques for analysis of this type of data, the collected data are
analysed; major conclusions are: The data from the Trimble receiver showed evidence of
autocorrelation in the offsets; the plough recorded a variance about three times that
recorded by the tractor.

File reproduced in accordance with the copyright policy of the publisher/author.