This thesis describes the design and creation of a system for stabilizing a camera mounted on a nano-UAV. A prototype voice coil actuator was produced, and Hall effect sensors were used to measure the angle of the actuator. The actuator and sensor is used to counter vibrations in the UAV. A circuit board used to control the actuator was designed and created, and a controller for the system was implemented. Testing showed that the controller was able to counter vibrations, but the bandwidth was low, only a few hertz. Wires connected to the actuator and sensor had significant impact on the system. They caused a spring/damper effect which meant more current was required, which again caused lower bandwidth due to saturation. On the other hand, the wires also lead to passive damping of higher frequencies. Unfortunately, there is a frequency band around 10 Hz where neither active nor passive control provides damping.

The idea about the CyberBike came to Jens G. Balchen - the founder of the Department of Engineering Cybernetics (ITK) at NTNU - in the 1980's. He wanted to make an unmanned autonomus bicycle, i.e. a bike that could run by itself. The idea was picked up by Amund Skavhaug, who started the CyberBike project in the late 80's. After being deffered for some years, the CyberBike has again gained some attention. This master's thesis is based on Hans Olav Loftum's and Lasse Bjermeland's theses at the spring 2006 and the autumn project of John A. Fossum the same year. The goal of the CyberBike project is to make the bike work as intended, i.e. as an autonomous unmanned bicycle. This thesis naturally share this goal, although the bike did not become able to take its first autonomous trip within the thesis' time frame. At the start of the work, the bike were already equipped with a suitcase of computational hardware on its baggage rack, a small QNX Neutrino OS image was installed on the industrial PC mounted in the suitcase, and drivers for the installed motors, tachometers and potmeters were written. An Inertial Measurement Unit (IMU) was intended to supply the control system with the necessary information about rotation, acceleration and position, and the unit was purchased for the purpose. Also a driver was written, but not properly tested. The IMU had to be installed and connected to the control system. The bike's control theory was developed, but had never been put into action outside computer simulations (due to the lack of acceleration measurements). The various tasks that had to be addressed emerged as the development process advanced. First, the IMU had to be connected to the system, by making a signal tranceiver circuit. A small printed circuit board was designed and laid out, mainly to include a MAX233CPP iC. Then the DB-9 serial connector on the bikes single board computer (Wafer-9371A) could be used to read the UART signal from the IMU as RS-232. Then some testing had to be done, and drivers updated. A better and more advanced IMU (referred to as the "MTi") was added to the project. This unit needed no signal converting circuitry, but driver development and testing still had to be done. To enhance the CyberBike's navigation opportunities, a GPS module was purchased. A signal transceiving circuit, similar to the one for the IMU, was made for this unit, as well as software to read out the measurements from the device. By the end of this thesis, no navigational algorithms are made, hence the GPS is currently not used, but available for future efforts made on this area. Some hardware related tasks was carried out, as connecting and implementing functionality to the pendulum limit switches, installation of a emergency stop switch and a power switch, purchasing and installation of two 12V batteries and a cooling fan. An operating system upgrade resulted in replacing the CyberBikePC's storage device, a compact flash card, with a mobile hard disk drive. Installation of a motor, for supplying torque to the rear wheel, included setup and tuning of a hardware based velocity controller in a Baldor TFM 060-06-01-3 servo module. However; this task is not to be considered as accomplished, due to some unsolved problems on the system I/O-card's output channels giving the motor controller card its reference voltage. A bike model and controller realized in Simulink was made by Bjermeland. Hence communicaton between Simulink and the device drivers had to be established, and this was realized by using S-functions and Real-Time Workshop. Finally the controller could be connected to the actual bike, but there was too little time left to explore this thoroughly, and make the system work properly. However, a foundation is laid for further development of the control strategy, hopefully storing a bright future for the CyberBike.

This study is a part of a renew and continuation of the SVEN work done in Sweden in the later 1970's. The SVEN hand was an on/off-controlled upper-limb prosthesis based on measured electromyographic (EMG) signals. Recently the SVEN methods have been revived in a cooperation by NTNU and UNB, Canada. The aim of this study is to further develop a practical porportional control system for a multifunction upper-limb prosthesis. This is based on a hypothesis that a simple and smooth proportional control system will be easier for the central nervous system to adapt to, compared to existing systems, and will thus provide increased functionality for the user. A protocol has been developed for the recording of EMG signals and VICON motion measurements in a laboratory. Suitable data sets have been recorded from three test subjects, and signal processing and three pattern recognition methods have been applied on these data sets to generate estimates of clinical angles. The pattern recognition methods tested were linear (LF) and quadratic (QF) mapping functions and multi-layer perceptron (MLP) network. The performance of these methods has been evaluated, compared and visualized. More testing is needed to find the best method, and the MLP network can be improved in several ways. To achieve better angle estimates that can be used for proportional control of prostheses, we wanted to use EMG signal features that are insensitive to amplitude changes due to variations in skin conductance. Qualitative and quantitative EMG signal features are described with this property as an important concern. The zero-crossings (ZC) feature has been tested as one of these, also in combination with the averaged absolute value (AAV). Although ZC did not always perform superior to AAV, it is likely that other features and combinations of these should be tested. Inclusion of other properties from the prosthesis, like elbow angle or measured pressure from the arm on the prosthesis, can also be included to improve the estimates. We now have a large, suitable data set from the laboratory, which can be used for further work on pattern recognition and multifunction proportional control of prostheses. There are also other applications for the methods developed. The final step will hopefully be implementation in a real prosthesis.

It would be of great value for the prosthesis industry to achieve an open standard for communication in upper limb prostheses. Cooperation between NTNU and the University of New Brunswick has resulted in a functional requirements specification for such a standard, SCIP(Standardised Communication Interface in Prostheses). The special challenges for communication in a prosthesis system are possible noisy environments, high demands for light weight, safety for the user and the fact that devices might be switched during operation. It was the purpose of this master thesis to make a design based upon those requirements. This was done by first choosing an existing bus standard, that would provide the lower levels of communication. CAN was chosen for this purpose. The next step of the design process was to transform the functional requirements into more specific technical requirements. This resulted in the definition of four types of nodes on the bus. These are bus controller, input controller, device controller and service controller. Their interactions called for the specification of several different message types, to support data exchange between the nodes. The result was a design that specifies node types, message types, variables like adresses, control strategies etc., state transition diagrams for the different node types and some message sequences. It also specifies the use of the CAN data-frame for all message types.

In this master's thesis, a ground station (GS) for the fixed-wing UAV: CyberSwan (CS), has been developed. The CS was designed for surveillance purposes, and two other master's theses deals with the work of making it autonomous (Høstmark (2007) and Bjørntvedt (2007)). Having a GS will make it possible to communicate with the CS in-flight, and present data and video from the CS through communication devices. The GS has been realised using LabVIEW development software from citet{labview}. A CS simulator was also developed in LabVIEW for test purposes. In addition was a Global Position System (GPS) receiver board, and a Radio Frequency (RF) communication board, developed. The GPS receiver was used to position the GS, and used as a source for position correction data. The RF communication board was developed for mounting in the CS and to be connected to its computer system to enable communication with the GS. The GS used a RF demo board for communication. A wireless camera was mounted on the CS for in-flight video surveillance, and a ultrasound ranging device was tested intended to be used in a autonomous landing situation. A hardware in the loop (HIL) test was performed to test the GS's communication capacity. Here the developed CS simulator was used, as the CS computer system was not completed (Bjørntvedt 2007). The test proved it possible to transfer a CS status message at 4 Hz, making the chosen communication device a good choice for the intended purpose.

This thesis takes upon itself to describe the medium of the radio channel for a WiMAX installation in a difficult environment. Furthermore it investigates the possibility of modifying the WiMAX standard in order to increase per- formance in such an environment. Channel estimators are crucial in optimizing throughput in WiMAX. Their performance is dependent on the amount of information in the pilot subcarriers of WiMAX. A way of adding information to the channel esti- mators without taking more of the bandwidth is to use a model in series with the data estimates. An already existing Kalman filter based algorithm for radio channel estimation is studied and compared to a WiMAX model. Modifications are suggested to fit the WiMAX signals. To extend a network fast in areas where there is little or no infrastructure to begin with is a task that can be performed by ad hoc network solutions. A statistical model for outage in ad hoc networks is studied and an evalua- tion on WiMAX’s capability of operating in an Ad hoc network structure is performed.

Shortly after birth, healthy infants exhibit so-called fidgety movements, while infants who later develop cerebral palsy (CP) lack these movements. General Movement Assessment (GMA) which is a clinical method, has proven its accuracy in detecting the absence (or presence) of fidgety movements, but for practical reasons, this method has not been adopted widely in the clinics. In order to create a similar but objective computer-based approach, Berg (2008) and Meinecke (2006) have studied discriminative features based on movement data collected from electromagnetic sensors and video. In this thesis, in addition to evaluation and comparison of previously introduced features, different classification methods have been applied to a suboptimal subset of these features. The results from linear and nonlinear separability analyses of features, confirm that dynamic features have better descriptive capabilities compared to statistically characterized features. Furthermore, it turns out that fidgety movements in the head (neck) and the arms show significant potential in distinguishing normal and abnormal infants, compared to signals from the trunk and the feet. The achieved results show 86% sensitivity and 90% specificity, which are highly acceptable, but this study needs further attendance before having any clinical usability. This study contains the first step of a typical medical research, meaning that the global (generalized) validity of the implemented methods are yet to be investigated, suppose that a representative selection (data) is available.

Healthy infants in the age group of 9 to 20 weeks post term have a distinct movement pattern called fidgety movements. The abscense of, or anomalies in these movements are indications that the infant may suffer from Cerebral Palsy. Data from video clips of infants have been extracted and used as features for classifying these movements as normal or abnormal. An earlier project used the concept of motiongrams for extracting the centroid of motion and quantity of motion of the image from each frame of the video clips. This project attempts to improve the dataset by extracting the same features using the concept of optical flow. Using the classification methods that yielded the best results from the motiongram project, the new dataset based on optical flow was constructed and tested for comparison. The image was as before divided into four quadrants with a circle of varying radius in the middle. Tracking how the centroid of motion moved between the separated areas of the image generated a transition probability matrix. By classifying on the entropy and variance of this matrix, a sensitivity of 90,0% and a specificity of 94,6% was reached, which was an improvement compared to the best classification result based on motiongrams with a sensitivity of 90,0% and specificity of 86,6%. The overall performance of both methods are around a sensitivity of 90% and a specificity of 70%. Using motiongrams and optical flow to extract features from video data for classification has turned out to be a promising approach for a simple and non-invasive objective tool for diagnosing infants with Celebral Palsy. The movement information in an optical flow field does however have a much higher potential for generating discriminative features than just as the simple representation a centroid of motion actually is.

Several feature extraction techniques, algorithms and toolkits are researched to investigate how speech recognition is performed. Spectrograms were found to be the simplest feature extraction techniques for visual representation of speech, and are explored and experimented with to see how phonemes are recognized. Hidden Markov models were found to be the best algorithms used for speech recognition. Hidden Markov model toolkit and Center for Spoken Language Understanding Toolkit, which are based on hidden Markov models, were not found to be suitable for the intentions of the thesis.

Two main types of formal methods have been investigated, formal specification and formal verification. Focus for formal verification has been on the concept of un-timed model checking. Some dominating formal specification languages, VDM and Z, and some prominent model checkers, FDR, Spin, and LTSA, have been learnt and presented. A tutorial for the formal verification tool Spin is created. The tutorial is example driven and describes the description language Promela and the verification methods available in Spin. Care has been taken to illustrate reasoning about the results from Spin. Topics discussed include the applicability and need for formal methods, the possible need for understanding the underlying theory, and considerations made in regards to creating the tutorial.