Agricultural Robotics

Abstract

A key goal of contemporary agriculture is to dramatically increase production of food, feed, fiber, and biofuel products in a sustainable fashion, facing the pressure of diminishing farm labor supply. Agricultural robots can accelerate plant breeding and advance data-driven precision farming with significantly reduced labor inputs by providing task-appropriate sensing and actuation at fine spatiotemporal resolutions. This article highlights the distinctive challenges imposed on ground robots by agricultural environments, which are characterized by wide variations in environmental conditions, diversity and complexity of plant canopy structures, and intraspecies biological variation of physical and chemical characteristics and responses to the environment. Existing approaches to address these challenges are presented, along with their limitations; possible future directions are also discussed. Two key observations are that biology (breeding) and horticultural practices can reduce variabilities at the source and that publicly available benchmark data sets are needed to increase perception robustness and performance despite variability.

Co-robotic harvest-aid platforms: Real-time control of picker lift heights to maximize harvesting efficiency

“…Harvesting robots recognize, locate, and detach fruits individually (Zhang et al, 2016). Some harvesting robots that handle fruits gently, without impacting fruit quality have been demonstrated (e.g., Bac et al, 2014, Silwal et al, 2017, Bogue, 2020, but are still at a pre-commercial stage, mainly because of low cost-effectiveness (high harvest cost), low picking speed compared to the manual harvesting, and inability to harvest a wide range of tree architectures (Vougioukas, 2019).…”

Row End Detection and Headland Turning Control for an Autonomous Banana-Picking Robot

“…A major challenge in improving headland turning navigation is the generation of a feasible turning trajectory based on route planning and motion planning, which numerical studies have mainly focused on [18]. In early studies, the main criteria, such as the minimum turning radius, the maximum lateral acceleration, and steering speed, were adopted using the Spline function, Bezier curve, etc., to generate the turning paths [19].…”

Robotic Tree-Fruit Harvesting With Telescoping Arms: A Study of Linear Fruit Reachability Under Geometric Constraints

“…The only remaining option, selective robotic harvesting, is still at a pre-commercial stage, despite active research for at least 30 years [20][21] [22]. To a large extent, this lack of progress can be attributed to robots' low cost-effectiveness (high harvest cost) and the inability to harvest a wide range of tree architectures [22].…”

“…The only remaining option, selective robotic harvesting, is still at a pre-commercial stage, despite active research for at least 30 years [20][21] [22]. To a large extent, this lack of progress can be attributed to robots' low cost-effectiveness (high harvest cost) and the inability to harvest a wide range of tree architectures [22]. The three parameters that most heavily influence harvest cost per robot arm are [23]: fruit picking efficiency (FPE), i.e., the ratio of fruits successfully picked over the total number of harvestable fruits; harvester purchase price (HPP); and fruit pick cycle time (PCT), i.e., the average number of seconds between successive fruit picks.…”

“…In general, the performance of robotic harvesters depends on the interrelationships among orchard layouts, tree canopy structures, and spatial fruit distributions with harvester mechanics [22]. In an effort to simplify the harvesting task, and increase fruit visibility and reachability, growers have been increasingly adopting high-density SNAP (Simple, Narrow, Accessible, and Productive) tree architectures [27].…”