This work presents Rakshak, a modular autonomous ground-surveillance rover designed for real-world perimeter security applications. The system integrates hybrid localization (GNSS/RTK and visual-inertial), dual-mode operation (autonomous patrol and manual override), and a hybrid edge–cloud perception pipeline for anomaly detection and biometric verification. The architecture emphasizes safety-critical local control, bandwidth-aware cloud offloading, and deployability under real-world constraints such as intermittent connectivity and heterogeneous terrain. The manuscript provides a detailed system design, implementation insights, and an evaluation framework for future benchmarking. The system has been physically deployed and tested in real-world environments, with accompanying architecture diagrams and operational visuals included in the paper. This work serves as a reference architecture for deployable AI-driven mobile surveillance systems in academic and industrial settings.
Salehmohamed et al. (Mon,) studied this question.