Speaker
Description
This study examines a real-time assistance concept designed to enhance operator decision-making and cognitive ergonomics in mechanized cut to length (CTL) forest harvesting. The concept integrates augmented reality (AR) visualization and mobile LiDAR sensing into a head-up hologram display (HUD) system that projects environmental information directly into the operator’s line of sight. Grounded in gamification principles, such as immediate feedback, improved situational awareness, and increased engagement, the system aims to support operators during complex work tasks that require simultaneous attention to machine control, timber quality assessment, and environmental considerations. A field experiment was carried out in eastern Finland, using a John Deere 1170G harvester equipped with an externally mounted Ouster OS0-128 LiDAR sensor and an in-cabin holographic HUD. The test involved the felling of 60 pre-marked Scots pine (Pinus sylvestris) trees under thinning conditions.
Real-time tree detection and diameter estimation were achieved through a LiDAR-based processing pipeline incorporating spherical projections and a YOLOv8s-seg instance segmentation model trained on 665 labelled images. Tree positions in three-dimensional space were derived by clustering point cloud data using DBSCAN, while diameter at breast height (DBH) was computed from circle fits applied to horizontal point cloud slices. Detected tree locations and DBH estimates were mapped to HUD pixel coordinates through geometric calibration and homographic transformations, enabling the operator to view holographically projected tree indicators aligned with the external forest environment.
The system demonstrated reliable performance in dynamic operating conditions, correctly projecting tree locations and preliminary DBH estimates for most of the test duration. However, intermittent detection errors and the need for periodic recalibration highlighted the challenges of maintaining stable alignment between the LiDAR sensor, HUD virtual image plane, and operator eye position. Operator workload was assessed using the NASA Task Load Index (NASA TLX), yielding a low mean score of 2.8, indicating minimal added cognitive strain. Post-test interviews further revealed generally positive attitudes toward the system’s potential for enhancing situational awareness and decision support.
Overall, the results demonstrate the feasibility of implementing holographic real-time environmental visualization in CTL harvesting but underscore the need for refinement in system alignment, data presentation, and ergonomic integration. Future development should focus on optimizing calibration procedures, improving visual clarity, and designing gamified elements that support operator performance without increasing cognitive load.
| Keywords | Ergonomic; Assistance; Harvesting; YOLOv8s-seg |
|---|
