Mahdi Nabiyouni

Slice-n-Swipe: A free-hand gesture user interface for 3D point cloud annotation

Three-dimensional point clouds are generated by devices such as laser scanners and depth cameras, but their output is a set of unstructured, unlabeled points. Many scenarios require users to identify parts of the point cloud through manual annotation. Inspired by the current generation of “natural user interface” technologies, we present Slice-n-Swipe, a technique for 3D point cloud annotation based on free-hand gesture input. The technique is based on a chefs knife metaphor, and uses progressive refinement to allow the user to specify the points of interest. We demonstrate the Slice-n-Swipe concept with a prototype using the Leap Motion Controller for free-hand gesture input and a 3D mouse for virtual camera control.

Poster: Designing effective travel techniques with bare-hand interaction

Emerging novel 3D interaction technologies allow precise tracking of bare hands and fingers, but due to the differences between these devices and traditional trackers, it is not clear how to design effective interaction techniques using these technologies. Using the Leap Motion Controller, we designed travel techniques with bare-hand interaction. We prototyped both unimanual and bimanual techniques using various metaphors (e.g., airplane and camera-in-hand), control mappings (position- vs. rate-control), camera movements (scene vs. camera dependent) and methods for speed control. Based on our experiences with these prototypes, we discuss the challenges and design issues for bare-hand interaction techniques. We present the results of a user study comparing the usability of five representative techniques for three travel tasks: absolute travel, naïve search and path following. We found that the limited workspace of the Leap caused movements with the camera-in-hand metaphor to be faster and less accurate, making it more effective for search but less effective for path following tasks. In addition, the Leaps ability to precisely track small finger movements benefited the usability of continuous speed control techniques.

Changes in RNA Splicing in Developing Soybean (Glycine max) Embryos.

Developing soybean seeds accumulate oils, proteins, and carbohydrates that are used as oxidizable substrates providing metabolic precursors and energy during seed germination. The accumulation of these storage compounds in developing seeds is highly regulated at multiple levels, including at transcriptional and post-transcriptional regulation. RNA sequencing was used to provide comprehensive information about transcriptional and post-transcriptional events that take place in developing soybean embryos. Bioinformatics analyses lead to the identification of different classes of alternatively spliced isoforms and corresponding changes in their levels on a global scale during soybean embryo development. Alternative splicing was associated with transcripts involved in various metabolic and developmental processes, including central carbon and nitrogen metabolism, induction of maturation and dormancy, and splicing itself. Detailed examination of selected RNA isoforms revealed alterations in individual domains that could result in changes in subcellular localization of the resulting proteins, protein-protein and enzyme-substrate interactions, and regulation of protein activities. Different isoforms may play an important role in regulating developmental and metabolic processes occurring at different stages in developing oilseed embryos.

A Highly Parallel Multi-class Pattern Classification on GPU

Multi-class pattern classification has a variety of applications and could be achieved using artificial neural networks (ANN). There are two major system architectures for using ANNs in multi-class pattern classification: using a single ANN and using multiple ANNs. Independent of what architecture is used, one of the main concerns of using ANNs is that with increasing number of pattern classes and training datasets, the training time will increase dramatically which renders the ANN unfeasible. In this paper, the vast computational power of Graphics Processing Units (GPUs) is utilized to mitigate this problem. Different architectures and different methods of feeding pattern classes are implemented in a GPU platform. Different methods have been proposed to achieve maximum parallelism and subsequently maximize throughput. Our implementation exceeds the state-of-the-art in literature in terms of speed and the accurate use of GPU resources. As a result, the proposed approachs run time is about 75% shorter than the previous approaches. In multi-ANN architecture, due to the inherent parallelism in the proposed implementation, the execution time of a system for a digit recognition application is reduced from seven hours in CPU to about 4 seconds in GPU.

Relative Effects of Real-world and Virtual-World Latency on an Augmented Reality Training Task: An AR Simulation Experiment

In Augmented Reality (AR), virtual objects and information are overlaid onto the user’s view of the physical world and can appear to become part of the real world. Accurate registration of virtual objects is a key requirement for an effective and natural AR system, but misregistration can break the illusion of virtual objects being part of the real world and disrupt immersion. End-to-end system latency severely impacts the quality of AR registration. In this research, we present a controlled study that aims at a deeper understanding of the effects of latency on virtual and real world imagery and its influences on task performance in an AR training task. We utilize an AR Simulation approach, in which an outdoor AR training task is simulated in a high-fidelity VR system. The real and augmented portions of the AR training scenarios are simulated in VR, affording us detailed control over a variety of immersion parameters and the ability to explore the effects of different types of simulated latency. We utilized a representative task inspired by outdoor AR military training systems to compare various AR system configurations, including optical see-through and video see-through setups with both matched and unmatched levels of real and virtual objects latency. Our findings indicate that users are able to perform significantly better when virtual and real-world latencies are matched (as in the case of simulated video-see-through AR with perfect augmentation-to-real-world registration). Unequal levels of latency led to reduction in performance, even when overall latency levels were lower compared to the matched case. The relative results hold up with increased overall latency.

A Taxonomy for Designing Walking-based Locomotion Techniques for Virtual Reality

Designers have yet to find a fully general and effective solution to solve the problem of walking in large or unlimited virtual environments. A detailed taxonomy of walking-based locomotion techniques would be beneficial to better understand, analyze, and design walking techniques for virtual reality (VR). We present a taxonomy that can help designers and researchers investigate the fundamental components of locomotion techniques. Researchers can create novel locomotion techniques by making choices from the components of this taxonomy, analyze and improve existing techniques, or perform experiments to evaluate locomotion techniques in detail using the organization we present.

Reduction of process variation effect on FPGAs using multiple configurations

In recent years, parameter variations present critical challenges for manufacturability and yield on integrated circuits. In this paper, a new method for improving the timing yield of field programmable gate array (FPGA) devices affected by random and systematic within-die variation is proposed. By selection of an appropriate configuration from a set of functionally equivalent configurations average critical path delay is reduced under conditions of large random and systematic variation considering spatial correlation. Compared to the previous approach which is limited to a fixed placement, our method improves timing yield by attempting several placements and routings without lengthy placement and routing phases to handle systematic variations and spatial correlation. The average critical path delay is reduced by 7% compared to the previous work over 20 MCNC benchmarks. General Terms: FPGA, process variation, placement, routing

Bookshelf and Bird: Enabling real walking in large VR spaces

We present two novel redirection techniques to enable real walking in large virtual environments (VEs) using only “room-scale” tracked spaces. The techniques, called Bookshelf and Bird, provide narrative-consistent redirection to keep the user inside the physical space, and require the user to walk to explore the VE. The underlying concept, called cell-based redirection, is to divide the virtual world into discrete cells that have the same size as the physical tracking space. The techniques then can redirect the users without altering the relationship between the user, the physical space, and the virtual cell. In this way, users can always access the entire cell using real walking.

Krinkle Cube: A Collaborative VR Game Using Natural Interaction

In this work we report our ongoing investigation on how natural interaction can improve player experience in collaborative virtual reality games. In particular we discuss bare-hand interaction and natural locomotion as means to reduce learning curve, improve immersion and allow non-verbal communication. Our findings concur with known benefits of natural locomotion, highlight the importance of good gesture design and bring new issues around facial expression and safety in co-located multiplayer games.

Interval Player: Designing a virtual musical instrument using in-air gestures

We present a novel virtual musical instrument interface based on in-air 3D gestures called the Interval Player. Instead of specifying absolute notes, the user specifies the melodic interval between the most recent note and the next one. The non-dominant hand is used to specify chords for harmony. We discuss the principles and rationale used in the design.