IntroductionMethodPlatformInterfaceConditionsParticipantsProcedureMetricsResults and DiscussionConclusionEnvironmental Factors Affecting Situation Awarenessin Unmanned Aerial VehiclesPrasanna Velagapudi∗, Sean Owens†, Paul Scerri‡and Katia Sycara§Carnegie Mellon University, Pittsburgh, PA, 15213, USAMichael Lewis¶University of Pittsburgh, Pittsburgh, PA, 15213, USAAs mini-UAVs become more capable and reliable, it is important to start looking at thefactors differentiating them from other classes of unmanned vehicles. One such factor is thephysical proximity of operators to the vehicle during deployment. Operators of these UAVsare often within sight of their vehicle, and share many environmental cues such as visuallandmarks. However, operating in the field also entails additional environmental stresses,such as less optimal use of computer equipment, variations in weather, and the physicaldemands of the terrain. In this paper, a pilot study is conducted to determine if any ofthese factors significantly impact situation awareness, by comparing operator performancein a visual identification task in a live field test with operators performing an identicaltask in a lab environment. Metric results suggest that performance is similar across thetwo conditions, but qualitative responses from participants suggest that the underlyingstrategies employed differ in the two conditions.I. IntroductionIn recent years, there has been a surge in the application of unmanned aerial systems to a wide varietyof problems. One particular class of vehicle that has been gaining favor in the community has been thefixed-wing mini-UAV. Small, low-cost, and portable, they are ideal for military and civilian applicationsranging from short-range surveillance1to environmental surveys2, 3to wilderness search-and-rescue.4One interesting aspect of these vehicles is that their deployment model differs greatly from their larger,longer range counterparts. Specifically, many scenarios for mini-UAV deployment involve short-range mis-sions, in which the UAV is deployed near (within a few miles of) the operator. In these cases terrain features,lighting, weather conditions, and even noises or sightings of the vehicle itself provide additional cues to hu-man operators that may generally assist in interpreting data from the UAV. On the other hand, additionalcognitive and physical costs are incurred by an operator making use of these cues, potentially increasingworkload and decreasing performance. For instance, the operator may be required to physically turn to facethe direction of the UAV, diverting attention away from a monitor showing a video feed, and requiring asubsequent mental context switch.In order to interpret data from a UAV and make decisions about actions based on the data, the operatormust maintain situation awareness.5Put simply, if the operator’s situation awareness is good, they canunderstand the context in which the UAV is operating and can interpret data coming from it more accuratelyand quickly. Like other forms of telepresence, UAV operators lack a number of potent cues, such as inertialfeedback and peripheral vision, thus degrading their overall situation awareness. In an extreme example,a soldier using a mini-UAV in a hostile tactical environment might detect an opponent in its video streambut take minutes to work out the position of the opponent relative to themselves. Situation awareness hasbeen well studied in the field of telerobotics, especially in applications pertaining to search and rescue,6–8∗Ph.D. Candidate, Robotics Institute, [email protected].†Research Programmer, Robotics Institute.‡Systems Scientist, Robotics Institute.§Research Professor, Robotics Institute.¶Professor, Department of Information Sciences & Telecommunications.1 of 7American Institute of Aeronautics and Astronauticsand the military.9, 10However, little work has looked at how the additional cues available to operators whenco-located with a UAV in some environment impact situation awareness. Intuitively, being able to see andhear the UAV and its position relative to features in the environment should improve situation awarenessbecause these extra cues are being provided in a very natural and familiar way. However, interpreting thesecues and relating them back to a map entails additional cognitive load. It is not known whether the benefitsof incorporating these cues outweigh the extra cognitive load when maintaining situation awareness.There is a significant body of literature dealing with the effects of various user interfaces on situationawareness on UAV platforms. Drury et al. presents a decomposition of the factors that constitute human-UAV situation awareness, breaking the problem into those of maintaining spatial relationships, knowledgeof vehicle intent and trust, and relevant environmental state.11In the case of mini-UAVs, these issuesare exacerbated by the very same characteristics that make them desirable. Low power and weight limitsconstrain sensing payloads to small sensors such as lightweight electro-optics, noisy GPS, and drifting inertialmeasurement. Operators are often limited to pose updates and streaming video of a small area of the ground.They must then use this information to identify targets, plan future trajectories and generally understand thesituation on the ground. Furthermore, portability constraints typically limit the user interface to computingplatforms that match the portability of the aircraft, such as laptops, PDAs, and embedded computing,limiting display and control hardware real-estate.Much previous work has also focused on optimizing the designs of human-computer interfaces withinthese constraints to maximize operator situation awareness. Goodrich and colleagues have developed numer-ous such interfaces, from 3D projective overlays12, 13to PDA clients.14Their work has focused mainly onpresenting information from the UAV to the operator in a more natural way, rather than examining or ex-ploiting the impact of the natural environment on task performance. In studies of UAV operator performancein the lab and in field trials for Wilderness Search and Rescue (WiSAR) tasks, Goodrich et al. acknowledgeand account for the additional workload of field operators by having operators perform secondary tasks inlab trials.15However, the potential beneficial impact of environmental cues and the actual difference inworkload between these conditions is not directly studied.The question