Search

We present ZeroEGGS, a neural network framework for speech-driven gesture generation with zero-shot style control by example. This means style can be controlled via only a short example motion clip, even for motion styles unseen during training. Our model uses a Variational framework to learn a style embedding, making it easy to modify style through latent space manipulation or blending and scaling of style embeddings. The probabilistic nature of our framework further enables the generation of a variety of outputs given the same input, addressing the stochastic nature of gesture motion. In a series of experiments, we first demonstrate the flexibility and generalizability of our model to new speakers and styles. In a user study, we then show that our model outperforms previous state-of-the-art techniques in naturalness of motion, appropriateness for speech, and style portrayal. Finally, we release a high-quality dataset of full-body gesture motion including fingers, with speech, spanning across 19 different styles.

The aim of this thesis was to accurately simulate the dynamic nature of known porous organic cage molecules, with a view to understanding the diffusion of different gases through their pore structures. Initial calculations showed that, due to their unique chemical structure, no ‘off-the-shelf’ force field (FF) was accurate enough to describe their dynamic motion. Therefore, a cage specific force field, (CSFF), was developed to be transferable across the first three cage systems, CC1-CC3. CSFF was subsequently used to rationalise the ‘on’/‘off’ porosity observed in two different polymorphs of CC1. A combination of computational simulations, including simulated surface area calculations, grand canonical Monte Carlo (GCMC) adsorption isotherms, and molecular dynamics (MD) simulations, for hydrogen and nitrogen in CC1α and CC1β, helped to confirm experimental results, as well as to provide further insight into why the polymorphism of CC1 alters the porosity of the molecule. In addition, CSFF was used to study the diffusion of a range of gases through crystalline CC3. Seven gases were chosen: hydrogen, nitrogen, carbon dioxide, methane, sulfur hexafluoride, krypton and xenon. A detailed understanding of the diffusivity within CC3 was accomplished by combining MD simulations with new methodologies and techniques, for example analysis of the dynamic connectivity. This helped to rationalise why CC3 showed good experimental uptake of gas, as well as highlighting potential separation capabilities. In summary, the development of CSFF has made it possible to simulate the diffusion of small gases through porous organic imine cages, and it has been shown that this diffusion is dependent on the relative size of the gas to the cage window, assuming that there is a suitable diffusion pathway. Using MD simulations, we have unlocked phenomena such as gas selectivity, rare-event hopping and the diffusion of gases to regions previously thought inaccessible. This has aided the rationalisation of existing experimental observations, and is a significant step forward for a priori prediction of porous organic cage systems, and their properties. This work has also led to new experiments that were prompted by my simulations. Finally, a new way to visualise the connectivity of a system has been introduced. This is achieved by monitoring how the surface area evolves with respect to time, during a MD simulation. This suggests how the pore channels of various systems, previously thought too small for gas adsorption, are actually suitable candidates for separations.

The Blackbeard instrument is a broadband radio receiver on the ALEXIS (Array of Low Energy X-ray Imaging Sensors) satellite. It detects and records transient VHF radio signals. During November and December of 1996, the Los Alamos Portable Pulser (LAPP) facility was used to transmit 31 broadband pulses to Blackbeard to evaluate the instrument's time tagging accuracy. LAPP firing times were used in conjunction with propagation delays to compute estimated times of arrival (ETOAs) for pulses reaching Blackbeard. ETOAs were compared to Blackbeard reported times of arrival (RTOAs), which were computed using information returned by Blackbeard and an algorithm presented in this paper. For the 31 pulser shots received by Blackbeard, the mean difference between ETOA and RTOA was 1.97 milliseconds, with RTOAs occurring later than ETOAs. The standard deviation of the difference was 0.43 milliseconds. As a result of the study, the algorithm used for accurate Blackbeard time tag studies has been modified to subtract 1.97 milliseconds from reported times of arrival. The 0.43 ms error standard deviation is now used to describe the uncertainty of Blackbeard time tags.

Efficient car shape design is a challenging problem in both the automotive industry and the computer animation/games industry. In this paper, we present a system to reconstruct the 3D car shape from a single 2D sketch image. To learn the correlation between 2D sketches and 3D cars, we propose a Variational Autoencoder deep neural network that takes a 2D sketch and generates a set of multi-view depth & mask images, which are more effective representation comparing to 3D mesh, and can be combined to form the 3D car shape. To ensure the volume and diversity of the training data, we propose a feature-preserving car mesh augmentation pipeline for data augmentation. Since deep learning has limited capacity to reconstruct fine-detail features, we propose a lazy learning approach that constructs a small subspace based on a few relevant car samples in the database. Due to the small size of such a subspace, fine details can be represented effectively with a small number of parameters. With a low-cost optimization process, a high-quality car with detailed features is created. Experimental results show that the system performs consistently to create highly realistic cars of substantially different shape and topology, with a very low computational cost.

Landmarks play a vital role in human wayfinding by providing the structure for mental spatial representations and indicating locations with which to orient. Less research effort has been allocated towards automated landmark identification in indoor environments despite a growing interest in indoor navigation in the scientific community. In this paper, we propose a computational framework to identify indoor landmarks that is based on a hierarchical multi-criteria decision model and grounded in theories of spatial cognition and human information processing. Our model of landmark salience is represented as a hierarchical integration process of low-level features derived from a three-part, higher-level, salience vector (i.e., cognitive, spatial, and subjective salience). We use a fuzzy hierarchical composite-weighted (objective and subjective) Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) to derive the rankings for identified objects at decision points (i.e., intersections). The top N objects are then selected and compared to a list of landmarks derived from an eye-tracking based virtual reality (VR) experiment. A substantial overlap of 79% was observed between these two lists. The proposed framework is capable of reliably and accurately detecting indoor landmarks, which can be employed in the development of landmark-based robot/autonomous agent motion and indoor guidance systems.

With the rapid increase in individuals participating in resistance training activities, the number of injuries pertaining to these activities has also grown just as aggressively. Diagnosing the causes of injuries and discomfort requires a large amount of resources from highly experienced physiotherapists. In this paper, we propose a new framework to analyse and visualize movement patterns during performance of four major compound lifts. The analysis generated will be used to efficiently determine whether the exercises are being performed correctly, ensuring anatomy remains within its functional range of motion, in order to prevent strain or discomfort that may lead to injury.