Using machine data to boost forest productivity

New Zealand forest harvesting operations are highly mechanised, with all stages of the harvest process, from felling to loading out, being achieved by machines. All the functions of these machines are controlled using the Controller Area Network (CAN) bus, a standard designed to enable efficient communication primarily between electronic control units.

As a result, a substantial amount of data sets are generated by these machines, including parameters such as engine speed, joystick movements and hydraulic pressures. Harnessing this data in real time offers significant benefits for operational monitoring and decision-making for operators, contractors and forest managers.

One of the projects in the FGR Automation & Robotics programme involves collecting machine data and transforming it into management information.

To do this, a system has been developed to tap into the harvesting machinery’s diagnostic ports to extract CAN bus data. Using Wi-Fi (Starlink), data is sent to the cloud where it is processed and then displayed in an interactive dashboard (see Figure 1). The dashboard enables real-time machine monitoring, as well as access to historical data from when the system was first installed and linked to GPS to display machine location and tracking over time. Several parameters can be visualised, including operating and idle time, fuel consumption and the terrain on which the machine is operating.

The significant amount of data gathered is only useful in the context of how the machine is completing tasks. To address this, a method was developed to use CAN bus data to predict machine activity during an operation. This method was then trialled on a John Deere forwarder. Results showed that the system achieved over 95 percent accuracy in predicting whether the forwarder was idling, loading, unloading and travelling loaded or unloaded. Once the machine activity is known, it becomes possible to dive deeper into actual operations.

For example, work cycles can be broken down into elements and then into individual movements. A method was developed based on an operator’s joystick movements to determine how many movements were required to complete a task. This approach allows standardised comparison of operator behaviour across different environments, operators and machines. This analysis will be useful for training new operators, identifying processes to increase or reduce workload and for enabling self-improvement through accurate and timely feedback to the operator.

By transforming CAN bus data into meaningful insights, decision-making, operation monitoring and training of new operators can be improved.

Harnessing machine data is an important first step towards achieving greater connectivity across the forest products supply chain.