Cognitive AI for Real-Time Cyber Threat Prediction in Autonomous Robotic Networks
Dr. M. Raju, Ms.S. Riddhi, M Srija · International Journal of Research Publication and Reviews · 2025
Autonomous robotic networks deployed in dynamic environments are increasingly vulnerable to sophisticated cyber threats.There is a growing need for predictive security mechanisms that can learn and adapt in real-time.This study introduces a cognitive AI model that combines reinforcement learning with attention-based transformer architectures to model adversarial behaviour in autonomous robotic systems.The model learns from sequential system states and threat indicators to estimate breach probabilities and suggest countermeasures.A testbed of autonomous delivery robots was used to simulate realistic threat scenarios including command injection and GPS spoofing.The model was trained over 5,000 episodes and compared with baseline static models.Results showed a 21% improvement in response accuracy and a 37% reduction in average threat handling time.This research highlights the potential of cognitive learning models in pre-empting cyber incidents and improving the robustness of robotic networks in real-world deployments.