Rapid Control Prototyping of Random Forest Model and its Performance Estimation by Simulation for Robotic Lawn Mower

This paper deals with Rapid Control Prototyping (RPC) of controller mounted on robotic lawn mower by using Model Based Development method. The controller optimizes the rotation speeds of two motors, one is for mowing the lawn grasses and another is for travelling the lawn mower by estimating whether the ground is with long lawn grasses or short lawn grasses, or without lawn grasses. A random forest model executes the estimation with high accuracy, and it is implemented on a Raspberry Pi by using the combination of Python and C++ language programs. The details of RPC are shown, and the time reduction for lawn mowing is estimated by simulation. The time reduction amounts to 27% compared with that of conventional controller against a real lawn field with the area of 31,170 m².