Techniques are disclosed for capturing and monitoring object motion in an AR environment. A first movement sequence may be received. The movement sequence may be an assigned movement routine. Image data of at least one target may be captured and the image data can be augmented with a training object to generate augmented training data. The training object can be caused to perform the first movement sequence. The motion of the at least one target can be recorded relative to the training object using the image data and one or more sensors deployed to the at least one target. A progression of therapeutic routines may be shown to the user depending on how the user's therapy is progressing. A second movement sequence can be received and the training object can be caused to perform the second movement sequence.