Robotic Tools for advanced agriculture Automation

摘要

The rapid progress in electronics and automation has led to a wide range of innovations in agriculture, especially in the field of automated driving or cooperative work processes. Some of the most popular examples are automatic steering systems or the assisted filling of transfer vehicles during the harvest process. Recently, also autonomous feeding and cleaning robots in cow barns were introduced in the market. The development of these systems is time and cost-intensive. Especially the huge amount of field tests is problematic in this regard. To solve problems in an early stage of development usually simulations are drawn up to test new concepts or functions before integrating them into a prototype. Nevertheless, in the area of vehicle automation and visual sensors a huge gap remains between simulation and field tests because so far many physical aspects of real world experiments have to be ignored. These restrictions are regarded to limited software and hardware capabilities or even due to the complexity of the system. This paper will show how to create simulations based on the Robot Operatic System (ROS) and the associated simulation environment Gazebo that can be used to approximate vehicle models and sensor data close to real world experiments. In detail it will be pointed out how to build up kinematic trees even of complex agricultural vehicles which are furthermore extended with visual sensors to develop automated functionalities.