Automatic recognition of work phases in cable yarding supported by sensor fusion

Abstract

Cable yarding is a semi-mechanized timber harvesting system that relies on human machine interaction where a considerable share of the work is done by forest workers. The system is used in mountain forests around the world. Automation of one or more functions could contribute to increased productivity, reduced physical workloads and improved safety conditions. This paper presents a method for sensor data fusion in order to automatically distinguish work phases using Partial Least Squares Discriminant Analysis (PLS-DA). The Robot Operation System (ROS) is implemented to allow for real-time data processing with a maximum latency of 0.16 s. Global Positioning System (GPS), Inertial Measurement Unit (IMU) and camera integration provided a robust solution for 78% correct process segmentation. These results provide a basis for further development from which there is a possibility of expanding this approach for semi-automation, remote control, and autonomous operation.