Pre-Training for Detection, Perception and Control of a Robotic Arm Mounted on a Legged/Wheeled Robot for Injecting Plants in the Plantation Farm

Khaja Ahmed Shaik

Keywords: Perception, Control, Robotic Vision, Robotics, BPN

TL;DR: Detection and Categorizing of Diseased plants (Banana & Coconut mostly) using BPN and Injecting the medicine in diseased plants with a 4 DOF Robotic Arm.

Abstract: This paper presents the pre-training for perception and control of a Robotic arm that is mounted on a top of Legged/Wheeled Robot (according to the users preference or environmental conditions) for injecting the medicine into the diseased plants in the plantation farms. It is a challenging and robust task for a human to roam in the plantation field and spot out the diseased plant and inject in its stem, instead according to the condition user/farmer can either select a legged robot (for confined space) or wheeled robot and mount the 4 DOF arm with a vision based sensor. The work in this paper into two important parts. Primarily pre-training the system to identify the diseased plant and then control of the manipulator to inject into the stem of the plant. The sensors used in this project are LiDAR for navigating itself between the plants and a vision based sensor (camera) for identifying the diseased plant.