This study presents an approach to predict the inverse kinematics of an industrial robotic manipulator using an Evolutionary Mating Algorithm (EMA) hybridized with Extreme Gradient Boosting (XGBoost). The proposed methodology aims to enhance the accuracy and efficiency of inverse kinematic predictions for the IRB 120 robotic manipulator. The dataset utilized for this research includes values for the robot’s joints (q1, q2, q3, q4, q5, q6) generated uniformly using Python's module, while the corresponding end-effector positions (x, y, z) are derived via the Denavit–Hartenberg parameters and forward kinematics equations. To validate the performance of the hybrid EMA-XGBoost, it has been compared against the Particle Swarm Optimization (PSO) integrated with XGBoost. The proposed EMA algorithm, inspired by natural mating behaviors and the Hardy–Weinberg principle, introduces new offspring generation and environmental adaptation mechanisms to optimize the search process. Performance metrics such as Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), maximum error, and the coefficient of determination (R2) were employed to evaluate the models for each output dimension (x, y, and z). The results demonstrate the efficacy of the hybrid EMA-XGBoost, which is the average R2 equal to 0.9567 compared to 0.9542 obtained by PSO-XGBoost, offering a promising solution for industrial applications in robotic manipulator control.