Intelligent Agriculture Robot for Tea Plantation Preservation - TeaBot

Tea cultivation stands out as one of the most significant export products that contribute to the Gross Domestic Product (GDP). In the previous decades, the substantial work-force gradually transitioned to different occupations. Compared to other crops, tea cultivation demands costly and meticulous upkeep. With a shortage of labor, maintaining tea estates grew tougher, leading to decreased yields. In reaction, estate owners transitioned to cultivating low-maintenance crops. The TeaBot is an advanced robot designed to replace human labor for watering and fertilizing vast tea estates. The TeaBot distinguishes itself by operating in rough outdoor terrain and infrastructure-free navigation in real tea plantations. Given that tea plants demand continuous hydration and nutrients for optimal crop yield, TeaBot plays a pivotal role in enhancing efficiency while reducing water and fertilizer wastage. Mainly four motor-powered wheels move accurately by translating linear and angular velocities using a precise motor control algorithm. An autonomous navigation algorithm was developed using two distinct approaches, which include deep learning-based computer vision and classical computer vision. The selection of the classical computer vision method was predicated upon its notable attributes, including high precision, minimal resource utilization, and optimal efficiency. A deep learning-based stem identification model was trained based on MobileNetV2 architecture to detect where the plant stem meets the ground for efficient hydration of individual plants. This lightweight model achieved 90% detection accuracy. The precise results of stem detection have made the liquid fertilization process more efficient.