Akademik Veri Yönetim Sistemi
Neural Computing and Applications, cilt.38, sa.4, 2026 (Scopus)
Chess is a game that is played between humans and adversaries of artificial intelligence (AI). In this paper, we propose a novel robotic chess system to counter the challenges presented by AI dominance, focusing on the integration of a Selective Compliance Articulated Robot Arm (SCARA) with advanced visual intelligence. By harnessing the capabilities of the SCARA robotic arm, we facilitate engaging and interactive gameplay experiences for players against software-based opponents. By integrating SCARA robot and the computer vision system, our approach enhances human-AI interaction, providing players with a tangible and immersive experience. We integrate Personalised Adaptive Learning (PAL) to leverage its capacity to bridge natural language understanding, problem decomposition, and external solving, thus enhancing AI-driven chess gameplay with increased sophistication and transparency. We adapt Large Language Model (LLM)-Planner framework for robotic game design, utilizing it to dynamically generate contextually grounded plans based on natural language instructions. We rigorously assess the performance of our robotic system through a comprehensive set of experiments conducted in both simulation and the real-world environment. We conducted experiments applying various manipulator control strategies in both simulated and real-world environments, alongside testing an obstacle avoidance algorithm and different object detection methods. The results showed more than 90% success rate in pick and place tasks with visual servoing control, while our obstacle avoidance algorithm reduced the average robot movement time by 15% and the use of linear actuators by 50%. YOLOv8 yielded the highest object detection performance with an mAP rate exceeding 97%. In general, our move correction algorithm effectively mitigated classification errors.