Click on “Download PDF” for the PDF version or on the title for the HTML version.

If you are not an ASABE member or if your employer has not arranged for access to the full-text, Click here for options.

Effort towards robotic apple harvesting in Washington State

Published by the American Society of Agricultural and Biological Engineers, St. Joseph, Michigan

Citation:  2016 ASABE Annual International Meeting  162460869.(doi:10.13031/aim.20162460869)
Authors:   Abhisesh Silwal, Joe Davidson , Manoj Karkee, Changki Mo, Qin Zhang, Karen Lewis
Keywords:   Apple identification, robotic harvesting, end–effector design, hand–picking dynamics, field evaluation

Abstract. Apple harvesting is not only labor intensive but also a time critical task requiring the rightly skilled workforce at the right time. The lack of mechanized or automated harvesting systems threatens the future of fresh market apple production because of the decreasing availability of farm labor. Over the last several decades, researchers have evaluated various types of mechanized and automated apple harvesting systems with limited successes. No commercially viable harvesting systems are available yet, primarily because of the challenges posed by the highly unstructured and biologically driven farming environment. This paper presents the novel approaches investigated at Washington State University to overcome these challenges in automated or robotic apple harvesting. First, a machine vision system capable of identifying apples in a naturally clustered and occluded conditions was developed using an over–the–row platform. The platform with artificial lighting provided a controlled imaging environment that minimized variability in lighting conditions and also provided capability for night time operation. Then, hand picking dynamics were studied to understand optimal picking patterns and forces required to detach apples from branches. Based on this study, a grasping end–effector was designed to meet requirements for robotic harvesting. The global vision system, robotic arm, and end–effector were then integrated and evaluated in a lab environment as a proof–of–concept followed by field evaluation in a commercial orchard in Prosser, WA. Results showed a huge potential for in–field automated robotic harvesting system capable of accurately identifying, localizing, and picking fruit at relative high speed. However, significant challenges for commercial implementation still remain. Future work to address these challenges are also discussed in this paper.

(Download PDF)    (Export to EndNotes)