<i>Hand-picking Dynamic Analysis for Robotic Agaricus Mushroom Harvesting</i>

摘要

<b><sc>Abstract.</sc></b> Mushrooms are picked manually one by one, which is time consuming and labor intensive. Robotic harvesting is an alternative method to address this issue. The purpose of this paper was to assess hand-picking methods as candidate techniques for implementation in a robotic end-effector designed to harvest Agaricus mushrooms. Conventional manual mushroom picking comprises a combination motion of twisting, bending, and lifting. To find an effective and simple picking method for robotic harvesting, a series of picking force and motion studies were conducted to compare among different picking methods. Four picking motions were considered, including conventional method, and three simplified methods, namely, bending, lifting, and twisting. The sensors used during picking included three force sensors and an inertial measurement unit. A damage test on the mushroom cap was also conducted to analyze the crop bruise level under different forces and acting duration. Experimental data were recorded for normal contact forces and mushroom rotation angles. The conventional picking method required the most orientation changes to detach a mushroom, which could generate complexity for designing a robotic end-effector with the equivalent dexterity. The test results indicated that bending was a simple and effective method for mushroom separation due to the least picking time and force needed compared to the other three patterns. Both lifting and twisting methods required a larger force on the index finger than on the thumb to detach mushrooms, which resulted in lower success pick rate and obvious bruise damage. The results also indicated that a bigger contacting force was the major reason for causing bruising, by comparing with the acting time. The outcomes could provide some guideline information for developing a robotic mushroom picking system for Agaricus mushroom harvesting.