Dynamic Sampling Algorithm for Agriculture-Monitoring Ground Robot

Abstract

We present the development and evaluation of a dynamic sampling algorithm for an agriculture-monitoring ground robot designed to locate insects in an agricultural field, where complete sampling of all plants is infeasible due to resource constraints. The algorithm utilizes real-time information to prioritise sampling at suspected points, locate hot spots and adapt sampling plans accordingly. A simulation environment was constructed to examine the algorithm's performance, and it was compared to two existing algorithms using Tetranychidae insect data from previous research. Sensitivity analyses reveals that the dynamic algorithm outperformed the others in all tested use cases, reaching 100 % detection approximately 3–5 days sooner when applied to small fields, and identifying 30 %–50 % more insects for larger fields. Its high detection percentages in small fields – 100 for a 1 ha field – decreased moderately with increasing field size to 80 % for a 10 ha field, seemingly irrespective of insect spread rate, which also barely affected insect detection. Doubling the time spent on each sample improved the results by 30–50 % on average in the first ten days, but in the following days the gap narrows. (