<![CDATA[Real-time control applications demand low-latency, reliable, and energy-efficient computational frameworks. Traditional cloud-centric architectures often fail to meet these requirements due to network-induced delays, unpredictable bandwidth, and limited adaptability under dynamic workloads. The integration of embedded intelligence within edge computing environments has emerged as a promising solution to enhance responsiveness, operational reliability, and system scalability. This research investigates edge computing architectures designed for embedded intelligence, aiming to optimize latency, throughput, and energy consumption in heterogeneous hardware configurations. Experimental and simulation-based methods were employed to evaluate performance across microcontrollers, FPGAs, and CPU/GPU nodes under varying workloads and network conditions. Data collection included latency measurements, throughput analysis, task completion times, and energy profiling. Inferential analyses, including correlation and regression models, quantified the relationship between computational capacity, responsiveness, and efficiency. A robotic manipulation case study further validated the practical application of the proposed architectures. Results indicate that adaptive, edge-enabled embedded intelligence significantly reduces latency to sub-10 millisecond levels, maintains high throughput, and ensures consistent task completion under dynamic conditions. Heterogeneous architectures outperform uniform deployments in both reliability and energy-performance balance. These findings demonstrate the feasibility and effectiveness of integrating embedded intelligence at the edge for real-time control. The study provides actionable guidance for designing scalable, robust, and energy-efficient intelligent control systems.]]>
