Having worked on some SLAM algorithms, it's important to make a prediction, then perform an action, then measure the results against the expected results. This

Having worked on some SLAM algorithms, it's important to make a prediction, then perform an action, then measure the results against the expected results. This process provides data and feedback about the accuracy of your model. If you ignore past predictions, you gain little insight and repeat the same erroneous process.