Faculty of Mechanical Engineering Niš, University of Niš

Faculty of Engineering, Catholic University of Croatia, Zagreb

Faculty of Electronic Engineering, Niš, Serbia

The paper is available at the following link: PDF

Abstract:

The rapid evolution of Industry 5.0 places humans back at the center of intelligent systems, demanding AI technologies that are not only powerful, but also explainable, trustworthy, and deployable in real-time on resource-constrained devices. This talk presents a unified perspective on edge-enabled artificial intelligence covering three emerging domains: sports analytics, collaborative robotics, and smart energy systems. We first discuss the development of vision-based deep learning models for on-device recognition of basketball events, human pose, and gesture dynamics, enabling real-time feedback and laying the foundation for AI-powered coaching assistants. We then transition to collaborative robotics, where robust perception of human motion and hand articulation becomes essential for safe and intuitive human–robot interaction in industrial environments. Since these tasks inherently unfold over time, we explore sequence modeling and time-series prediction, including LSTM- based and hybrid approaches used for forecasting human motion as well as for large-scale applications such as predictive control in district heating systems. Finally, we emphasize the role of Explainable AI (XAI) in bridging the gap between model performance, decision transparency, and operational trust- particularly in high-impact engineering systems. The talk highlights how edge computing, human-centric design, and XAI jointly shape the next generation of intelligent small-scale systems.

Keywords:

Artificial Intelligence, Machine Vision, Edge Computing, eXplainable Artificial Intelligence.

REFERENCES:

[1] Ćirić, I., Simonović, M., Pavlović, M., Cvetković, S., Mašinska Vizija u Mehatronici, 1st ed., 2025, ISBN: 978-86-6055-198-8.
[2] Chen, N., et al. (2023). Real-time vision-based human hand motion tracking and grasping for a robotic manipulator with a soft hand. Progress in Canadian Mechanical Engineering, CSME - CFD-SC2023 May 28-31, 2023, Sherbrooke, QC, Canada
[3] Ćirić, I., Ivačko, N., Ćojbašić, Ž., Pavlović, M., Mitić, D., Lalić, S., Jevtić, D., "Comparison of Simple Object Recognition Algorithms for Robot Vision", Proc. XVI Int. SAUM Conference on Systems, Automatic Control and Measurements, Niš, Serbia, Nov. 17–18, 2022.
[4] Ivačko, N., Ćirić, I., Cojbašić, Ž., Milošević, M., Jevtić, D., "Machine Learning Based Hardware Accelerated Cobot Grasping in the Food Industry", Proc. 11th Int. Conf. on Electrical, Electronic and Computing Engineering (IcETRAN), Niš, Serbia, pp. 1–5, 2024.
[5] Ivačko, N., Stojiljković, D., Jevtić, D., Simonović, M., Ćirić, I., "AI-Based Robot Vision Algorithm for Fruit Ripeness Classification", Proc. 17th Int. Conf. on Accomplishments in Mechanical and Industrial Engineering (DEMI 2025), Banja Luka, Bosnia and Herzegovina, pp. 441–448, 2025.
[6] Ivačko, N., Ćirić, I., Stojiljković, D., Ćojbašić, Ž., Jevtić, D., "A Cognitive Robotics Approach for Manipulation of Freeform Objects Using CNN-Based Perception and Soft-Gripping", in Advances in Service and Industrial Robotics: RAAD 2025, Mechanisms and Machine Science, vol. 190, Cham, Switzerland: Springer, pp. 46–54, 2025, doi: 10.1007/978-3-032- 02106-9.
[7] Ćirić, I.; Ivačko, N.; Milić, M.; Ristić, P.; Krstić, D. Computer Vision-Driven Framework for IoT-Enabled Basketball Score Tracking. Computers 2025, 14, 469. https://doi.org/10.3390/computers14110469
[8] Ćirić, I., Milošević, M., Ivačko, N., Pavlović, M., Krstić, D., "Intelligent Computer Vision System for Score Detection in Basketball", FACTA UNIVERSITATIS, Series: Automatic Control and Robotics, vol. 22, no. 2, pp. 075–085, 2023.
[9] Milić, M., Ristić, P., Ćirić, I., Ivačko, N., Krstić, D., "Edge Computing for Sports Analytics: AI-Powered Basketball Score Detection on Raspberry Pi 5", Proc. IEEE 34th Int. Conf. on Microelectronics (MIEL), 2025.
[10] Hu, M., Hu, Q., "Design of Basketball Game Image Acquisition and Processing System Based on Machine Vision and Image Processor", Microprocessors and Microsystems, vol. 82, no. 1, 2021, doi: 10.1016/j.micpro.2021.103904.
[11] Ćirić, I., Pavlović, M., Ivačko, N., Tomov, P., Nikolić, V., "Autonomous Train Operation Obstacle Detection Based on Night Vision System", Acta Polytechnica Hungarica, vol. 23, no. 1, pp. 307–327, Jan. 2026, doi: 10.12700/APH.23.1.2026.1.18.
[12] Cvetkovic, S., Savic, N., Ciric, I., "Deep Transfer Learning Approach for Robust Hand Detection", Intelligent Automation & Soft Computing, vol. 36, no. 1, pp. 967–979, 2022, doi: 10.32604/iasc.2023.032526.
[13] Dai, J. Advancements in Deep Learning for Visual Object Tracking, Appl. Comput. Eng., 82, 130–136, 2024.
[14] Feng, H. et al. Benchmark Analysis of YOLO Performance on Edge Intelligence Devices, Cryptography, 6, 16, 2022
[15] Zdravković, M., Ignjatović, M., Ćirić, I., "Explainable Heat Demand Forecasting for the Novel Control Strategies of District Heating Systems", Annual Reviews in Control, vol. 53, pp. 405–413, 2023, doi: 10.1016/j.arcontrol.2022.03.009.
[16] Ćirić, I., Tasić, M., Ignjatović, M., Stojiljković, D., Ćirić, M., "Machine Learning Approach for Energy Consumption Estimation in District Heating Systems", Innovative Mechanical Engineering, vol. 3, pp. 31–47, ISSN 2812-9229.
[17] Ćirić, M., Predić, B., Stojanović, D., Ćirić, I., "Single and Multiple Separate LSTM Neural Networks for Multiple Output Feature Purchase Prediction", Electronics, vol. 12, no. 12, 2616, 2023, doi: 10.3390/electronics12122616.
[18] Tang, P., Li, Z., Wang, X., Liu, X., & Mou, P. (2025). Time series data augmentation for energy consumption data based on improved TimeGAN. Sensors, 25(2), 493. https://doi.org/10.3390/s25020493
[19] Neubauer, A., Brandt, S., and M. Kriegel. 2025. Explainable multi-step heating load forecasting: Using SHAP values and temporal attention mechanisms for enhanced interpretability. Energy and AI 20:100480.