Improving Robot Image Perception Design Using Artificial Intelligence

In this current decade, there are numerous growth has been identified in science and technology. Traditional robots functions and its performance are frequently failed in fulfilling the people's production and life needs, whereas the robot development has evolved into increasingly intelligent as well as networked. For these, the key technology for mobile robot location as well as its navigation, environment perception technology remains a priority and interesting area of study in the robotic fields. Methods for image processing have had an important effect on the vast majority of robotics as well as automation in industry. Real-time approaches are commonly utilized in complicated automation tasks, assist in making decision or effective guidance of machinery and robots, whereas post-processing is usually applied for retrospective technology and its process evaluation. In addition, a human pose detection system is built using Deep Learning (DL) as well as robotics whereas each part is recommended. Ultimately, the overall system efficiency is verified. The outcomes indicate that once the precision approaches High-Resolution Net (HRNet-32). The model's inferences performance is boosted by around 2.4 times, while its parameters are made compressed over approximately 67%. The results evaluation with existing Convolutional Neural Network (CNN) has illustrate the method's effectiveness and the posture detection method in robot’s human has ability in resolving human posture detection effectively in fitness activities.