Design and Implementation of a Control System for an Autonomous Reforestation Machine Using Finite State Machines

Abstract

Reforestation is performed after the final felling as an important and often law-mandated step to ensure that wood production is sustainable. In Sweden alone, over 400 millions seedlings are planted annually. This work is physically demanding and the quality is uneven. Therefore, automatic production systems are under research and development. A necessary effort in this endeavor is presented in this paper: the development and evaluation of a mission supervisor utilized to control the mission and behavior of a full-scale autonomous forest regeneration machine tested in realistic environments. The mission supervisor is implemented in the Robot Operating System framework using a finite state machine package called SMACH. A terrain machine built as a research platform with an added full-scale forwarder crane is used as a base machine. First, we describe the scenario in which planting is conducted, whereupon we develop the composite tasks required as states. A simplified simulator then enables an intermediate step before field experiments. The system is implemented and operated in real time on a full-scale machine. Results show that the developed SMACH mission supervisor can be used as a sound basis for an autonomous forest regeneration machine and the chosen communication solution between different systems works well. The simulations show good agreement with the experiments. The results also show that crane movements take 70% of the machine time, emphasizing the importance of limiting crane movement, improving the actuator movement speed and integrating the composite solutions. Further development with a holistic approach is required before the concept can reach the prototype level.