A review on vision-based humanoid robot technologies: Exploring agricultural applications

다국어 초록 (Multilingual Abstract)

Recent advances in artificial intelligence, sensor integration, and robotic control technologies have led to a rapid expansion of robotic applications across various sectors. Among these, humanoid robots—characterized by human-like forms and functions—are increasingly utilized not only in industrial settings but also in healthcare, disaster response, and service industries. Within this technological trend, the agricultural sector is gaining attention as a potential frontier for humanoid robot deployment, representing the next phase of automation. Agriculture involves highly complex and unpredictable environments, including seasonal variability, where conventional fixed or wheeled automation systems often fall short. Humanoid robots, with their human-like range of motion and sensory feedback capabilities, possess the potential to flexibly adapt to such challenging conditions. This paper aims to analyze the technological potential of humanoid robots in agricultural environments. To this end, it first reviews the structural features and application cases of representative humanoid robots. It then examines technical developments centered on vision-based crop recognition and environmental perception, control algorithms including reinforcement and unsupervised learning, and multi-sensor fusion systems. Furthermore, it discusses the impact and applicability of next-generation cognitive control systems, such as the emerging Vision-Language-Action (VLA) models, in agricultural robotics. This study provides a comprehensive outlook on the utilization of humanoid robots in agriculture and proposes a technological foundation for advancing precision agriculture and future human–robot collaboration systems.