Order Picking is one of the most important tasks in modern warehouses. Since most work is still done manually, new methods to improve efficiency of the task are being researched. While the currently most used approaches Pick-by-Paper and Pick-by-Light are either prone to error or only scalable with high costs, other methods are considered. These methods include Pick-by-Vision systems based on Augmented Reality although these systems mostly rely on head-mounted displays. In order to evaluate a new method, we developed OrderPickAR which uses an order picking cart as well as projected user interfaces. OrderPickAR is part of the motionEAP project of the University of Stuttgart and relies on in-situ projection as well as motion recognition to guide the user and present feedback. The intuitive feedback provided by the in-situ projection as well as the motion recognition gives OrderPickAR the chance to effectivly eliminate errors while lowering the task completion time. With the use of a mobile workstation we also address the scalability of OrderPickAR. Since the developement is not sufficiant, we also conducted a study in which we compared OrderPickAR to currently used approaches. In addition we included a Pick-by-Vision approach developed in a related project by Sebastian Pickl. We analysed and compared different error types as well as the task completion time.