Surrogate models based on convolutional neural networks (CNNs) for computational fluid dynamics (CFD) simulations are investigated. In particular, the flow field inside two-dimensional channels with a sudden expansion and an obstacle is predicted using an image representation of the geometry as the input. Generative adversarial neural networks (GANs) have been shown to excel at such image-to-image translation tasks. This motivates the focus of this work on investigating the specific effect of adversarial training on model performance. Numerical results show that the overall accuracy of the GANs is generally lower compared to an identical generator model trained directly on the ground truth using an L1 data loss. On the other hand, GAN predictions are often visually more convincing and exhibit a lower continuity residual.

Numerous critical manual teleoperation tasks, such the control of the refueling boom during aerial refueling, require human controllers to accurately manipulate objects in the depth dimension, i.e., aligned with the viewing direction. To better understand the intricacies of depth control tasks and to be able to better support human controllers in such tasks, a cybernetic analysis of human control behavior in stereoscopic vision-enhanced depth control tasks would be a valuable extension of the current state-of-the-art in manual control research. This paper presents the initial findings of a human-in-the-loop experiment in which participants performed an abstract pursuit tracking task in which multisine target and disturbance forcing functions were used to facilitate cybernetic analysis of the measured control behavior. In terms of depth perception (i.e., perspective, viewing distance), the task was modeled after an aerial refueling scenario. Participants performed the pursuit tracking task for a reference "flat-plane" condition (task axis aligned with vertical screen axis) and depth tracking tasks either without stereoscopic cues, with natural stereoscopic vision, and with amplified hyperstereoscopic vision. Overall, the results of the experiment showed that participants achieved degraded task performance and less effective control dynamics in depth tracking tasks compared the the reference "flat-plane" condition. However, in line with earlier research on aerial refueling operator support systems, increased strength of the stereoscopic vision enhancements is found to enable much improved performance and increased human control gains.