Author : Raj Kumar Gudivaka, Rajya Lakshmi Gudivaka and Thanjaivadivel M

Autonomous robotics demands real-time decision making and adaptive signal processing for effective operating in changing environments. This work introduces Edge AI-based IoT signal processing platform that maximizes autonomy, reduces latency and enhances computational efficiency in robotics systems. Current approaches including Centralized Cloud-Based AI, Federated Learning and Homomorphic Encryption have high latency (65 ms), high CPU usage (80%) and data leakage threats (12.4%). Suggested system combines lightweight neural networks, adaptive signal filtering and real-time sensor fusion for accurate feature extraction and decision-making. Fuzzy logic-based adaptive control system enhances system responsiveness during uncertain situations. Experimentation proves there is high performance gain where suggested system attains accuracy of 0.95 in object detection, lowering latency to 28 ms and decreasing energy consumption rate by 37% over cloud-based AI. It improves noise reduction effectiveness by 45 percent providing accurate sensor data processing. Privacy is obtained through privacy-preserved AI methods like Split Learning, neutralizing threats related to data leakage in decentralized AI systems. Findings validate that proposed framework supports efficient, secure and real-time processing for IoT-based autonomous robotic systems positioning it as perfect solution for future smart automation scenarios. Research contributes to the field of AI-powered automation, IoT-integrated intelligent systems and future-generation autonomous robots driving developments in real-time perception, control and safety.

Keywords: Edge AI, IoT Signal Processing, Autonomous Robotics, Fuzzy Logic Control, Split Learning and YOLOv8