The document discusses the development of robust adaptive inverse models using Bacterial Foraging Optimization (BFO). Three robust cost functions are used to develop robust inverse models that are less impacted by outliers in training data. BFO is used to iteratively minimize the robust cost functions. Simulation results show that the inverse model trained with the Wilcoxon norm-based cost function provides the best performance and is most robust to outliers, outperforming models trained with other cost functions or standard squared error. The approach develops inverse models that are more accurate and reliable even in the presence of outliers in the training data.