Background. Common manufactured depth sensors generate depth images that humans normally obtain from their eyes and hands. Various designs converting spatial data into sound have been recently proposed, speculating on their applicability as sensory substitution devices (SSDs). Objective. We tested such a design as a travel aid in a navigation task. Methods. Our portable device (MeloSee) converted 2D array of a depth image into melody in real-time. Distance from the sensor was translated into sound intensity, stereo-modulated laterally, and the pitch represented verticality. Twenty-one blindfolded young adults navigated along four different paths during two sessions separated by one-week interval. In some instances, a dual task required them to recognize a temporal pattern applied through a tactile vibrator while they navigated. Results. Participants learnt how to use the system on both new paths and on those they had already navigated from. Based on travel time and errors, performance improved from one week to the next. The dual task was achieved successfully, slightly affecting but not preventing effective navigation. Conclusions. The use of Kinect-type sensors to implement SSDs is promising, but it is restricted to indoor use and it is inefficient on too short range.

No abstract

International audienceNowadays, it is possible to build a multi-GPU supercomputer, well suited for implementation of digital signal processing algorithms, for a few thousand dollars. However, to achieve the highest performance with this kind of architecture, the programmer has to focus on inter-processor communications, tasks synchronization. In this paper, we propose a high level programming model based on a data flow graph (DFG) allowing an efficient implementation of digital signal processing applications on a multi-GPU computer cluster. This DFG-based design flow abstracts the underlying architecture. We focus particularly on the efficient implementation of communications by automating computation-communication overlap, which can lead to significant speedups as shown in the presented benchmark. The approach is validated on three experiments: a multi-host multi-gpu benchmark, a 3D granulometry application developed for research on materials and an application for computing visual saliency maps