Robot for Optimized and Autonomous Mission-Enhancement Responses (ROAMER)

Abstract

This system presents the development of an advanced mobile manipulator system designed for banana plantations, focusing on task performance in highly unstructured and cluttered environments. The system consists of five primary modules: mobile robot platform, navigation and mapping, disease detection, communication, and the decision support system. The mobile robot incorporates a rocker-bogie mechanism for unstructured terrain traversal and a 5-DOF robotic arm which employs Computed Torque Control (CTC) and Gross Control Fine Motion (GCFM) control strategies for precise and efficient spraying on banana plants. The mission planning subsystem schedules and executes tasks based on user input, with the system enabling seamless communication via the Robot Operating System (ROS). The robot’s computer vision module uses a multi-view deep-learning model based on VGG16 for disease detection and classification. The system processes the gathered data for reporting, allowing users to access insights through a web-based interface. This integrated approach facilitates autonomous operation in banana plantations, supporting efficient plant health monitoring and task execution.