Comparison of Shallow Neural Network with Random Forest Algorithm in Estimating Lawn Grass Lengths for Robotic Lawn Mowers

This paper states the accuracy comparison of shallow neural network (SNN) with random forest algorithm in the estimation of lawn grass lengths for robotic lawn mowers. This relates to Digital Twin and Virtual Twin of Hybrid Twin approach when controlling the autonomous driving of robotic lawn mower. In its autonomous operating, the length estimation of lawn grasses or of such ground conditions as dirt, gravel, or concrete, etc., are important for precisely controlling the rotation speed of motor and moreover, the estimation must be fast. Therefore, an SNN is constructed, and its performances are compared with those of random forest algorithm through some experiments. The accuracy of SNN is 0.7 point better than that of random forest algorithm and the training time is very fast, that is, less than 0.4 seconds. The SNN is being further improved from the view point of application range and performances.