混合式視覺伺服控制在農作採收之應用

Abstract

本論文「混合式視覺伺服控制在農作採收之應用」旨在利用不同視覺伺服控制方法開發可應用於溫室農作物採收之農業機器人。研究中利用一台深度攝影機即可獲取玉女番茄在三維空間中的姿態，可針對實際中不同角度生長方向之番茄進行視覺伺服，也探討了不同視覺伺服控制方法，包含基於位置之視覺伺服(Position-Based Visual Servo, PBVS)與基於圖像之視覺伺服(Image-Based Visual Servo, IBVS)，並提出了基於模糊動態控制參數之混合式視覺伺服控制(Hybrid Visual Servoing Control, HVSC)，探討不同視覺伺服控制方法之特性，並應用在實際農作採收中。結果顯示本研究所開發之混合式視覺伺服控制對玉女番茄之平均採收時間為9.40s/per，平均採收成功率為96.25%。This paper "Application of Hybrid Visual Servo Control in Agricultural Harvesting " aims to develop agricultural robots that can be applied to greenhouse crop harvesting by using different visual servo control methods. In the research, a depth camera can be used to obtain the posture of the tomato in three-dimensional space, and visual servo control can be carried out for the tomato growing at different angles in practice. Different visual servo control methods are also discussed, including thePosition-Based Visual Servo (PBVS), Image-Based Visual Servo (IBVS) and the proposed.Hybrid Visual Servoing Control (HVSC) based on the fuzzy dynamic control parameters. Discuss the characteristics of different visual servo control methods were discussed, and then applied to actual harvesting. The results show that the hybrid visual servo control developed in this research has an average harvesting time of 9.40s/per and an average harvesting success rate of 96.25% for Cherry tomato.