Dexterous hand technologies have emerged as a critical frontier in robotics, driven by growing demands for human-like manipulation in complex and dynamic environments. This review explores how advancements in computer science—particularly in intelligent control, sensor fusion, and machine learning—have transformed the design and deployment of dexterous robotic hands. We analyze three major integration pathways: data-driven control algorithms that improve real-time adaptability, application-specific solutions in domains such as healthcare and manufacturing, and multimodal perception frameworks that enhance manipulation precision through the fusion of tactile, visual, and proprioceptive data. The review also identifies key challenges, including perception latency, data scarcity, limited generalization, and deployment barriers. Finally, we propose future research directions involving context-aware learning, simulation-based training, lightweight architectures, and interdisciplinary collaboration. The findings underline the importance of computational intelligence in enabling robotic hands to achieve robust, safe, and flexible performance in real-world settings.
Di Li (2025) studied this question.