Search

Research on information processing in neuronal networks may lead to a better understanding of many brain diseases and disorders and could inform potential future treatments. It may also lead to the development of more intelligent electronic technology. To understand how the brain processes information we need to study the patterns of action potentials (APs) in a neuronal network over the long-term. The aim of this project, therefore, is to develop an optical system to both detect and stimulate APs in a small neuronal network. A closed-loop system has been developed that consists of a signal detection tool (a camera system), a signal analysis tool (a real-time computer) and a signal delivery tool (a spatially modulated light source). The conventional techniques for measuring APs, using patch clamp electrodes or multi electrode arrays for example, are limited by the spatial extent, or the spatial resolution, that can be achieved. Alternatively, optical imaging systems can be used to monitor APs. Imaging systems can achieve high-resolution across a wide field-of-view. To record network-wide neuronal activity a high- speed camera is necessary. Ideally imaging needs to be performed continuously over long periods of time. However, continuous high-speed imaging is typically not possible using commercially available systems due to restrictions in bandwidth.

Human-machine cooperative or co-robotics has been recognized as the next generation of robotics. In contrast to current systems that use limited-reasoning strategies or address problems in narrow contexts, new co-robot systems will be characterized by their flexibility, resourcefulness, varied modeling or reasoning approaches, and use of real-world data in real time, demonstrating a level of intelligence and adaptability seen in humans and animals. The research I focused is in the two sub-field of co-robotics: teleoperation and telepresence. We firstly explore the ways of teleoperation using mixed reality techniques. I proposed a new type of display: hybrid-reality display (HRD) system, which utilizes commodity projection device to project captured video frame onto 3D replica of the actual target surface. It provides a direct alignment between the frame of reference for the human subject and that of the displayed image. The advantage of this approach lies in the fact that no wearing device needed for the users, providing minimal intrusiveness and accommodating users eyes during focusing. The field-of-view is also significantly increased. From a user-centered design standpoint, the HRD is motivated by teleoperation accidents, incidents, and user research in military reconnaissance etc. Teleoperation in these environments is compromised by the Keyhole Effect, which results from the limited field of view of reference. The technique contribution of the proposed HRD system is the multi-system calibration which mainly involves motion sensor, projector, cameras and robotic arm. Due to the purpose of the system, the accuracy of calibration should also be restricted within millimeter level. The followed up research of HRD is focused on high accuracy 3D reconstruction of the replica via commodity devices for better alignment of video frame. Conventional 3D scanner lacks either depth resolution or be very expensive. We proposed a structured light scanning based 3D sensing system with accuracy within 1 millimeter while robust to global illumination and surface reflection. Extensive user study prove the performance of our proposed algorithm. In order to compensate the unsynchronization between the local station and remote station due to latency introduced during data sensing and communication, 1-step-ahead predictive control algorithm is presented. The latency between human control and robot movement can be formulated as a linear equation group with a smooth coefficient ranging from 0 to 1. This predictive control algorithm can be further formulated by optimizing a cost function. We then explore the aspect of telepresence. Many hardware designs have been developed to allow a camera to be placed optically directly behind the screen. The purpose of such setups is to enable two-way video teleconferencing that maintains eye-contact. However, the image from the see-through camera usually exhibits a number of imaging artifacts such as low signal to noise ratio, incorrect color balance, and lost of details. Thus we develop a novel image enhancement framework that utilizes an auxiliary color+depth camera that is mounted on the side of the screen. By fusing the information from both cameras, we are able to significantly improve the quality of the see-through image. Experimental results have demonstrated that our fusion method compares favorably against traditional image enhancement/warping methods that uses only a single image.

When analyzing time-to-event data, informative dropout due to competing risks is one prac- tical aspect that researchers should take into account. If we fail to account for the association between the event of interest and informative dropouts, unknown amplitude bias may be en- countered when identifying the effects of potential risk factors related to time to the main cause of failure. A joint modeling approach of time to the main event and time to the competing events is proposed, to capture the dependence between the main event and the informative dropout due to competing risks via a set of random terms. Two fundamental likelihood functions with different structures of the random terms are provided, which may be combined in practice. We used three methods to optimize the corresponding likelihood functions in order to estimate the unknown covariate effects: Gaussian quadrature method, Bayesian Markov Chain Monte Carlo method, and hierarchical likelihood method. Four bias reduction correction methods for the h-likelihood estimation approach are explored. These methods were aimed to improve the accuracy of parameter estimation. The performances of the three methods were compared via simulations. We applied proposed methods to identify risk factors for dementia. Time-to-event data have been widely investigated from clinical trials and from obser- vational studies. The proposed joint modeling method is significantly meaningful to public health research because informative dropout commonly exists for the time-to-event data. Methods that have been currently used either fail to adjust for the association between the main event and the informative dropout due to competing events or the methods used to adjust for the association are not easy to implement. In this dissertation, we showed that the proposed joint modeling approach provides less bias estimates on the effect of a risk factor and has fairly straightforward implementation, which will lead to benefits for medical research.

As a special subset of ad hoc networks, airborne networks aim to provide efficient network access for airborne and ground assets in a tactical environment. Conventional ad hoc routing protocols face some difficulties in such networks. First, significant overhead may be generated due to the high node mobility and dramatic topology changes. Second, temporary link failure may abort the delivery of a packet in some intermediate router. In this thesis, we propose a cluster-based reactive routing protocol to alleviate these problems. Our solution takes advantage of mesh routers installed in unmanned aerial vehicles or aircraft capable of hovering, when such airborne assets are available. As those mesh points usually have relatively stable connections among themselves, they play the role of cluster heads, forming a hierarchical routing structure. A simple self-organizing rule is introduced in cluster management to limit the cluster control overhead and route discovery flooding. In addition, a disruption tolerant mechanism (DTM) is deployed in the routing protocol to increase resilience to temporary link or node failure. The DTM utilizes the location, bearing and speed information provided by each node and intelligently maintains a buffer of packets that cannot be immediately delivered. If a temporary link failure occurs in the intermediate router during delivery, the packet is then buffered in that router up to a maximum time-to-live. The DTM also keeps track of link changes and tries to deliver the message as soon as a new path toward the destination is found. If the buffered messages are about to time out and the destination is still unreachable, the DTM still makes an effort to deliver the packet to another router with higher probability of eventually reaching the destination. This thesis also presents an implementation of the proposed solution in the ns-2 network simulator. The conventional Ad hoc On-Demand Distance Vector (AODV) routing protocol is adopted as the base model in the implementation. A mesh router model is programmed with two wireless interfaces. One of the interfaces is utilized to exchange routing information and packets with cluster members; the other is used to communicate with other mesh routers. This model is then installed on top of the AODV routing protocol and forms the hierarchical routing structure. The traditional AODV messages, including RREQ, RREP and HELLO, and routing tables are modified to support additional location information. Finally, the DTM is programmed and added to the AODV buffer management. The objective of this research is to use a mesh structure and DTM to improve the reliability and performance of airborne networks. The metrics of throughput and routing overhead are taken into consideration. The simulation results demonstrate that the proposed solution satisfies our research objectives. It achieves better performance than the conventional AODV, but introduces little overhead. The mesh structure can effectively adapt to high mobility, dynamic topology and different routing capabilities. The DTM provides a sophisticated way to maintain the buffer and mitigates the impact of intermittent links.