Canopy density estimation in perennial horticulture crops using 3D spinning lidar SLAM

We propose a novel, canopy density estimation solution using 3D ray cloud representation for perennial horticultural crops at the field scale. To attain high spatial and temporal fidelity in conditions, we application of continuous-time SLAM (Simultaneous Localisation Mapping) to spinning lidar payload (AgScan3D) mounted on moving farm vehicle. The AgScan3D data is processed through Continuous-Time algorithm into globally registered cloud. global canonical format (a digital twin) from which can compare vineyard snapshots over multiple times within season across seasons. Then, rows are automatically extracted novel calculation performed estimate maximum likelihood densities vineyard. This combination twinning, together with accurate extraction structure information, allows entire vineyards be analysed compared, growing year year. proposed method evaluated both simulation experiments. Field experiments were four sites, varied vine management, two seasons 64 collection campaigns, resulting total traversal 160 kilometres, 42.4 scanned hectares vines combined approximately 93,000 vines. Our show repeatability 3.8% (Relative RMSE) per panel, acquisition speeds 5-6 km/h, under half standard deviation estimated when compared an industry gap-fraction based solution. code datasets available https://github.com/csiro-robotics/agscan3d.