Abstract
This paper explores the intersection of deep learning and robotics, focusing on the development and implementation of structured deep visual models for improving the perception and control of robotic systems in manipulation tasks. The integration of convolution neural networks (CNNs), recurrent neural networks (RNNs), and attention mechanisms plays a pivotal role in enabling robots to efficiently interpret visual data and make informed decisions in complex and dynamic environments. The presented research contributes to the ongoing efforts in bridging the gap between perception and action in robotics, paving the way for more robust and intelligent manipulation capabilities in diverse and challenging environments. The insights gained from this study offer valuable guidance for researchers, engineers, and practitioners working on the forefront of advancing robotic systems through the integration of deep learning techniques. Structured deep visual models contribute to improving a robot's ability to perceive and understand its environment. This includes advancements in object recognition, pose estimation, and scene understanding, allowing robots to interact with their surroundings more intelligently. The integration of deep learning in robot manipulation empowers robots to operate with greater autonomy. Structured visual models enable robots to make informed decisions based on visual data, reducing the need for explicit programming and enhancing adaptability in dynamic environments. The acceptability of alternatives is gauged by comparing them to the average response. Utilizing the EDAS method, this assessment determines the most advantageous answer by considering both the average evaluation and its deviation from the mean solution. According to the analysis, EDAS favors solutions closer to the ideal solution, while penalizing those with negative deviations, indicating a preference for options that closely align with the ideal. From the result Graph Neural Networks (GNNs) is got the first rank where as Generative Adversarial Networks (GANs) is having the lowest rank.