Real-Time Threat Detection and Autonomous Decision-Making Using Deep Learning and A* Path Planning

This paper has presented a modular, browser-deployable autonomous surveillance system integrating three computational modules: a YOLOv8-based real-time object detector, an A* grid path planner operating on detection-derived obstacle maps, and a hierarchical priority decision engine. The system is implemented entirely in Python within a Streamlit web application, enabling deployment on any machine with a webcam and standard Python dependencies without specialized runtime infrastructure. Experimental evaluation across controlled indoor scenarios demonstrated overall detection macro-recall of 86.2%, path planning success of 97.1% across diverse obstacle densities, zero hard obstacle path violations, and a decision engine accuracy of 91.8% against human-annotated ground truth. The system achieves 8.3 frames per second on CPU hardware and 27.6 frames per second with GPU acceleration. The principal contribution of this work is the demonstration that a coherent, functionally capable perception-planning-decision pipeline for indoor surveillance can be assembled from open-source components and deployed as a standard web application without requiring robotics middleware, cloud infrastructure, or specialized hardware. The identified limitations regarding weapon coverage, temporal memory, and decision uncertainty quantification provide a clear roadmap for subsequent development toward operationally deployable capability.

[1]  J. Redmon, S. Divvala, R. Girshick, and A. Farhadi, "You Only Look Once: Unified, Real-Time Object Detection," Proc. IEEE CVPR, Las Vegas, NV, USA, June 2016, pp. 779-788. [2]   Ultralytics, "YOLOv8: A New State-of-the-Art Computer Vision Model," Ultralytics Documentation, 2023. [Online]. Available: https://docs.ultralytics.com [3]  T.-Y. Lin, M. Maire, S. Belongie, J. Hays, P. Perona, D. Ramanan, P. Dollar, and C. L. Zitnick, "Microsoft COCO: Common Objects in Context," Proc. ECCV, Zurich, Switzerland, September 2014, pp. 740-755. [4]  P. E. Hart, N. J. Nilsson, and B. Raphael, "A Formal Basis for the Heuristic Determination of Minimum Cost Paths," IEEE Transactions on Systems Science and Cybernetics, vol. 4, no. 2, pp. 100-107, July 1968. [5]  S. Koenig and M. Likhachev, "D* Lite," Proc. 18th AAAI Conference on Artificial Intelligence, Edmonton, Canada, August 2002, pp. 476-483. [6] K. Daniel, A. Nash, S. Koenig, and A. Felner, "Theta*: Any-Angle Path Planning on Grids," Journal of Artificial Intelligence Research, vol. 39, pp. 533-579, 2010. [7] R. A. Brooks, "A Robust Layered Control System for a Mobile Robot," IEEE Journal of Robotics and Automation, vol. 2, no. 1, pp. 14-23, March 1986. [8]  M. Quigley, K. Conley, B. Gerkey, J. Faust, T. Foote, J. Leibs, R. Wheeler, and A. Ng, "ROS: An Open-Source Robot Operating System," Proc. ICRA Workshop on Open Source Software, Kobe, Japan, May 2009. [9]  M. Shaik and J. Kandasamy, "Multi-Source Fake Review Detection using Deep Learning (LSTM)," ISJEM, vol. 05, no. 05, May 2026. [10]  A. Dosovitskiy et al., "An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale," Proc. ICLR, Vienna, Austria, May 2021.