According to one embodiment of the invention, a computerized method for predicting posture of a living object includes storing a plurality of posturing rules for a respective activity in a storage location, receiving a start posture of a first living object for the activity, receiving a location of a target for the activity, dividing the activity into a plurality of movements based on the start posture and the location of the target, and determining an end posture of the first living object for the activity. Determining the end posture includes applying at least one of the posturing rules for each of the movements and checking at least one of a plurality of constraints associated with each of the movements.