FORWARD: Dataset of a forwarder operating in rough terrain

TL;DR

FORWARD provides an open, centimeter-accurate, multimodal dataset of a large forest forwarder operating on rough terrain, combining telematics, IMUs, a 360° camera, vibration sensing, airborne LiDAR terrain data, drone imagery, and StanForD production logs across two Swedish sites. The dataset includes 18 hours of annotated work and a suite of controlled driving scenarios to study traversability, perception, and autonomous control, with an accompanying analysis toolkit and data processing scripts. It enables detailed traversability analyses, simulator calibration, and benchmarking of automation solutions in forestry, with potential impacts on efficiency, safety, and environmental impact. Overall, FORWARD fills a gap in open data for automated forestry robotics by providing richly annotated, synchronized multimodal field data suitable for data-driven and physics-based modeling and validation.

Abstract

We present FORWARD, a high-resolution multimodal dataset of a cut-to-length forwarder operating in rough terrain on two harvest sites in the middle part of Sweden. The forwarder is a large Komatsu model equipped with vehicle telematics sensors, including global positioning via satellite navigation, movement sensors, accelerometers, and engine sensors. The vehicle was additionally equipped with cameras, operator vibration sensors, and multiple IMUs. The data includes event time logs recorded at 5 Hz of driving speed, fuel consumption, vehicle position with centimeter accuracy, and crane use while the vehicle operates in forest areas, aerially laser-scanned with a resolution of around 1500 points per square meter. Production log files (StanForD standard) with time-stamped machine events, extensive video material, and terrain data in various formats are included as well. About 18 hours of regular wood extraction work during three days is annotated from 360-video material into individual work elements and included in the dataset. We also include scenario specifications of conducted experiments on forest roads and in terrain. Scenarios include repeatedly driving the same routes with and without steel tracks, different load weights, and different target driving speeds. The dataset is intended for developing models and algorithms for trafficability, perception, and autonomous control of forest machines using artificial intelligence, simulation, and experiments on physical testbeds. In part, we focus on forwarders traversing terrain, avoiding or handling obstacles, and loading or unloading logs, with consideration for efficiency, fuel consumption, safety, and environmental impact. Other benefits of the open dataset include the ability to explore auto-generation and calibration of forestry machine simulators and automation scenario descriptions using the data recorded in the field.

Figures (22)

• Figure 1: Interactive plot for visualizing and navigating the time series data. The user can zoom in and out, and select preferred time intervals to show experiments, video files, and exact time interval in the terminal (purple box in the figure) for further analysis. Blue lines, orange areas, and green areas indicate experiments, raw 360°-video filenames, and lightweight stitched video filenames, respectively.
• Figure 2: Visualization of the three steps from unclassified point cloud (a), via classified point cloud (b), to interpolated terrain model (c) for part of the Björsjö site, including our test circuit marked in red. Colors from blue to red reflect low to high elevation.
• Figure 3: Overview of the testsites with terrain elevation maps obtained from the airborne laser scanning overlayed by orthomosaic photos from drone images and machine paths from the GNSS positioning data.
• Figure 4: 360° camera mounted on the forwarder cabin (left) and IMU mounted on the wheels (right).
• Figure 5: Forwarder during experiments on flat, sandy surface in old quarry. Empty, 10 tons, and 20 tons load weight.