Abstract
Power tillers are widely used in small-scale agriculture, particularly in developing countries, where they provide a cost-effective alternative to conventional tractors. However, their manual operation remains labor intensive and susceptible to inefficiencies in field conditions. This study presents the design and implementation of an autonomous power tiller integrating mechatronics and machine learning for enhanced operational efficiency. The system employs ultrasonic sensors for obstacle detection and avoidance, while an Arduino microcontroller governs control functions. Additionally, machine learning techniques are utilized to predict tilling performance based on soil conditions, terrain, and environmental factors. A supervised learning model was trained using historical data to optimize tiller path planning and soil engagement efficiency. Results from experimental trials and simulations indicate that the autonomous power tiller significantly reduces manual effort while maintaining effective tilling performance. The integration of predictive analytics enhances adaptability, making it a promising solution for sustainable mechanized farming. Future research should explore additional sensor integration, real-time data analytics, and networked autonomous agricultural robotics