In this paper, a new Internet of Things (IoT) smart irrigation system is proposed that empowers an autonomous vehicle with predictive analytics to enforce efficient water management in farming. The suggested “Smart Agro-Mobile” system offers a combination of real-time sensing, autonomous mobility, and machine learning to maximize agricultural irrigation systems. It has three-layered architecture consisting of sensing (rain sensor, temperature, soil moisture), processing (Arduino Uno WiFi and Raspberry Pi 3), and cloud analytics layers. The mobile platform, which consists of 12 V LiPo battery, 6 V mini pump, and ultrasonic sensors, travels autonomously in fields to collect environmental information and make precise irrigation. Enhanced Long Short-Term Memory (LSTM) networks, supplemented with dropout layers and attention mechanisms, predict soil moisture content and precipitation possibilities with 92% accuracy, enabling preventive irrigation scheduling. Experimental validation confirms significant improvement over the base method with 40% water conserved while maintaining optimal soil moisture level 98% of the time. Its applicability to all crops and soil conditions is assured by its modularity and its adaptive threshold principle. With removal of the limitations of static sensor networks and incorporation of weather volatility in irrigation decisions, this solution offers an inexpensive, effective approach to precision agriculture. Merging IoT with autonomous mobility is a paradigm shift in agricultural water management, promising tremendous gains for resource-limited small-scale farmers.