Evaluating the Effectiveness of Mixed Reality Simulations for Developing UAV Systems

The development cycle of an Unmanned Aerial Vehicle (UAV) system can be long and challenging. Mixed Reality (MR) simulations can reduce cost, duration and risk of the development process by enabling the replacement of expensive, dangerous, or not yet fully developed components with virtual counterparts. However, there has been little validation of such hybrid simulation methods in practical robot applications. This paper evaluates the use of MR simulations for prototyping a UAV system to be deployed for a dairy farming monitoring task. We show that by augmenting the robot's sensing with a virtual moving cow using an extensible Augmented Reality (AR) tracking technique, MR simulations could help to provide efficient testing and identify improvements to the UAV controller. User study findings reveal the importance of both virtual and MR simulations to robot development, with MR simulations helping developers transition to development in a more physical environment.