DEVELOPMENT OF ROBOTIC BLACK PEPPER HARVESTING SYSTEM

Abstract

Black pepper is a perennial crop and one of India's most economically significant spices. It has a high commercial value in the market all around the world. Its fruit is harvested, dried, and powdered for many cuisines and processed for many value-added products. Black pepper is a flowering vine growing on supporting stakes. The berries turn from green to red on maturity and are harvested when it starts to turn red. For achieving good quality and good-sized pepper, it should be harvested at its correct maturity stage. Generally, black pepper spikes were harvested manually by climbing on supporting trees using bamboo poles. It is a tedious task because there are chances of falling from ladders while harvesting and also causes some musculoskeletal diseases to the labours. For their time saving and heavy work intensity, farmers harvest almost all the fruits in a range of maturity along with the real matured ones. This practice eventually affects the crop yield and quality. Through robotic harvesting, black pepper spikes can be harvested at correct maturity and also helps to overcome the difficulties faced by the labours. The main functions of robotic harvesting are identification, plucking, depositing, and controlling. KAU developed a machine vision system with the camera as sensor, Raspberry pi 4 model B as the processor, and LCD as the display unit to identify matured black pepper spikes. The programing code was written in python language, and the Tensorflow-faster RCNN platform was used for the detection. Hence, a robotic black pepper harvesting system was developed in the present study, and its performance evaluation was carried out. The physical properties of black pepper relevant to design and develop a robotic black pepper harvesting system were determined. The developed robotic black pepper harvesting system consists of a machine vision system to identify matured black pepper spikes, a manipulator with 2 DOF, an end-effector with 1 DOF, and a control unit. Servo motors actuated the shoulder and elbow joints of the manipulator and the cutting blades. Shear-type cutting was employed for detaching pepper spikes from the pepper vine. The entire system was controlled by the microprocessor Raspberry pi 4 Model B. For controlling the servo motors, the library RPi.GPIO was installed on raspberry pi, and the programming code waswritten in python language. Two lead-acid batteries with a voltage of 12 V and a current 9Ah were connected in parallel to power the entire system. The overall dimension of the developed unit was 59 × 18 × 162 cm, and it weighs 2.1 kg. The performance evaluation parameters of the machine vision system viz., sensitivity, specificity, and accuracy were respectively as 85 %, 77 %, and 82 % in Karimunda variety and 84 %, 77 %, and 82 % in Panniyur 1 variety. Time taken for detection is 0.43 seconds. Also, the capacity of the developed robotic black pepper harvesting system is 3.5 kg h-1 and 562 spikes h-1 in the Karimunda variety, whereas 4.6 kg h -1 and 683 spikes h -1 in Panniyur 1 variety. The effectiveness index, time taken for the entire operation, harvesting loss, and drying loss was 81%, 6.6 seconds, 4.9 %, and 39 % in the Karimunda variety and 82 %, 6.3 seconds, 7%, and 66 % in Panniyur 1 variety respectively. The system takes 0.18 seconds for a single cut for both varieties; it was fixed in the program. A study was also carried out for manual harvesting and found that manual harvesting has a capacity of 1052 spikes h-1 and 6.3 kg h -1 in the Karimunda variety and 1654 spikes h -1 and 10.8 kg h -1 in the Panniyur 1 variety, which is higher than robotic harvesting. The effectiveness index of the manual harvesting was 40% in Karimunda and 38 % in Panniyur 1, which is lower than robotic harvesting. The harvesting loss and drying loss of manual harvesting are 15.3 % and 56 % in Karimunda and 17.5 % and 81 % in Panniyur 1, which is higher than robotic harvesting. It was statistically verified and found a significant difference between manual and robotic harvesting in terms of capacity, effectiveness index, harvesting loss, and drying loss at a 5 % level of significance.