Multi-sensor simulation method for outdoor plant phenotyping based on autonomous field robots.

Abstract

Based on an existing field robot for outdoor plant phenotyping a simulation environment has been built up for multi-sensor simulation. The development environment Microsoft Robotics Studio, typically used for 3D robot navigation purposes, has been adapted to this application by creating services representing sensors such as 3D-Time-ofFlight cameras, light curtains, color cameras, GPS or incremental encoders. As the real robot the virtual sensor data are stored in a data base and can be analyzed with sensor fusion algorithms. First comparisons of real and simulated data show good overall agreement, in particular an artificial plant and its corresponding virtual counterpart have been compared. There is a high potential of the new method for plant phenotyping purposes as well as for other field robot applications.