This paper describes an Integrated Autonomous Robot and Intelligent Agricultural IoT system, designed to bridge the gap between macro-environmental monitoring provided by large-scale sensor networks and the fine-grained assessments of plant health. A decision framework combines both modalities to enable precise fertigation and climate control based on detected stress and local microclimate analysis. This aim is achieved by an innovative system that continuously monitors the plant health inside the greenhouse using (i) an autonomous vehicle with SLAM with 2D/3D path mapping, (ii) an IoT Node to collect/predict available/missed data, and (iii) a Robot Arm to operate actions at plant level.The robot settings utilize mission-driven scheduling and acquire optimal pathways through learning algorithms. Its arm, designed according to Denavit-Hartenberg convention, performs opeartions, such as cultivating ripe tomatoes, removing deemed/diseased one, cutting off a dead leaf or broken root, weeding around the crop. The robotic agent uses a CV TensorRT-based pipeline to provide physiological stress at the vegetative level.Our IoT Node gathers plant microclimate data, such as: Temperature, Humidity, air quality, Light Intensity, and Soil Salinity and camera pictures related to its health state. Any missing or faulty readings are reconstructed using GRU Random Forest, LSTM, PLSTM, and Transformer algorithms, reaching R² scores of up to 0.999 and a mean absolute error of less than 0.02. A deep learning suite (YOLOv11-26, RF-DETR, ResNet) conducts real-time plant’s health assessment, with an accuracy reaching 87,5% in crops classification and 97,5% in leave’s health recognition, along with a wound segmentation mAP@0.5 of 92%.