The quality of a teleoperation system is decreased by time delays in the communication channel. Delays as low as a few hundred milliseconds between commanding an action and getting the visual feedback reduce the operator's performance. Predictive displays have proven their suitability to compensate for these delays, but at the expense of image quality when using computer-generated images. A photorealistic predictive display is presented that closes the feedback loop locally at the operator's side of a telepresence system. Photorealism is achieved using delayed camera images for texturing the predicted scene. Consumer graphics hardware is not only used for rendering but also for hardware-accelerated texture extraction. To allow concurrent access to model data, a multibuffer model structure is presented. A model of the teleoperator's environment is automatically acquired and updated by image processing techniques using a stereo camera as the only sensor.

No abstract

No abstract

The growth of compute-intensive applications causes an increasing demand for computing resources in both data centers and underlying networks. To satisfy these computing and networking demands, hardware-accelerated computing with GPUs and FPGAs becomes more and more prevalent. However, current approaches of accelerated Virtual Network Functions (VNF), typically based on PCIe accelerator cards, suffer from many superfluous data transfers between the memory of hosts and accelerator devices.In this work we propose "host bypassing" for chained hardware-accelerated network functions, which reduces memory copying to a minimum, which increases throughput and reduces latency.