U Cihazlarda Ger ek Zamanli 3B Nokta Bulut Segmentasyonu

As autonomous systems evolve, they have begun to interact more with their environment. To execute their operational cycles flawlessly, it is critical for these systems to perceive their surroundings at high processing speeds. Additionally, such systems may have constraints related to volume, weight, and energy consumption. Our study focuses on executing high-speed perception cycles that meet these constraints and enable environmental interaction. The real-time operational performance of these cycles was examined on a selected edge device. Among the evaluation metrics, we included the "number of processed images per unit of energy consumed,"which is not commonly found in similar studies. The results clearly indicate that edge devices hold significant potential for real-time point cloud segmentation.