Development of smart industrial model using sensor network

Abstract

This thesis explores using reinforcement learning to optimize newlinemobile robot and UAV performance in simulated smart industrial newlineenvironments. The research utilizes the MATLAB framework to train newlinealgorithms (PPO, TRPO, Policy Gradient, DQN) for improving key newlineparameters like speed, accuracy, and battery life. The evaluation focused on newlinehow each algorithm impacted these parameters. PPO achieved consistent newlineimprovements in speed, payload capacity, and accuracy for both mobile newlinerobots and UAVs. TRPO also showed strong performance across various newlineparameters. PGRL and DQN exhibited mixed results, with DQN excelling in newlinereliability. This study highlights the potential of reinforcement learning, newlineparticularly PPO, for optimizing performance in smart industrial applications. newlineFuture work should refine algorithms, expand parameter optimization, and newlineincorporate real-world data for even more effective autonomous systems in newlineindustrial settings. newlineThe study focuses on enhancing mobile robot and UAV newlineperformance in smart industries. Using reinforcement learning algorithms, newlinethe goal is to optimize key performance parameters such as speed, accuracy, newlineand reliability in simulated environments. By utilizing reinforcement learning newlinealgorithms through the MATLAB framework, the study aims to optimize key newlineperformance parameters of these autonomous systems. Firstly, the work newlinefocuses on the selection and implementation of reinforcement learning newlinealgorithms leveraging the capabilities of the MATLAB platform newline