Canopy-Based Monte Carlo Localization in Orchards Using Top-View Imagery

Localization of ground mobile robots in orchards is a complex problem which is yet to be fully addressed. The typical localization approaches are not adjusted to the characteristics of the orchard environment, especially the homogeneous scenery. To alleviate these difficulties, we propose to use top-view images of the orchard acquired in real-time. The top-view observation of the orchard provides a unique signature of every tree formed by the shape of its canopy. This practically changes the homogeneity premise in orchards and paves the way for addressing the kidnapped robot problem. Using computer vision techniques, we build a virtual canopies laser scan around the ground robot which is generated from low-altitude top-view video streams. We apply Monte Carlo Localization on this virtual scan to localize the robot against a high-altitude top-view snapshot image which is used as a map. The suggested approach is examined in numerous offline experiments conducted on data acquired in real orchards and is compared against a typical simulated approach which relies on ground-level trunk observations. The canopy-based approach demonstrated better performance in all measures, including convergence to centimeter-level accuracy.