The use of tilting bodies on railway vehicles is becoming increasingly widespread: a number of well-established services using tilt technology already exist around the world, and will appear again in the UK over the next year or so. The motivation for tilting railway vehicles is that they give a cost-effective means of achieving a substantial reduction in journey time by increasing the vehicle speed during curves. Early tilt controller designs were based upon local vehicle measurements, however at that time this approach did not prove very successful. Nowadays most European manufacturers use the so called ‘precedence' control scheme, utilising measurements from precedent vehicles to achieve ‘precedence' information. However, achieving a satisfactory local tilt control strategy is still an important research target because of the system simplifications and more straightforward failure detection. The thesis describes a comprehensive study of tilt control, and its aim is to employ advanced control techniques - based upon practical sensors - with the particular objective of identifying effective strategies which can be applied to each vehicle independently, i.e. without using precedence control. The sensors employed for control design are in particular mounted on the vehicle passenger coach. Most of the work has been undertaken using Matlab, and this has included a proper assessment of the ride quality issues.

Narrow tilting vehicles offer an opportunity to tackle both traffic congestion and carbon emissions having a small footprint, low weight and small frontal area. Their narrow width requires that they tilt into corners in order to maintain stability; this may be achieved by means of an automated tilt control system. A three-wheeled tilting vehicle prototype, known as the Compact Low Emission Vehicle for uRban transport (CLEVER), was constructed at the University of Bath in 2006. The vehicle was equipped with a direct tilt control system in which a pair of hydraulic actuators applied a moment between the cabin and a non-tilting base. This tilt control system provided satisfactory steady state performance but limited transient stability. High tilt rate demands associated with rapid steering inputs would lead to large tilting moments being applied to the non-tilting rear engine module; this, combined with the engine module’s own propensity to roll out of the bend, could cause the inside wheel to lift and the vehicle to capsize. This thesis details the implementation of a Steering Direct Tilt Control (SDTC) system, whereby the front wheel steer angle is used to generate some of the tilting moment, on the prototype CLEVER Vehicle. Simulation and experimental results are presented which show a 40% reduction in load transfer across the rear axle during a transient ramp steer manoeuvre. The influence of the SDTC system, and associated steer angle alteration, on the vehicle trajectory is considered. A human driver is found to be capable of adapting their steer inputs such that they can follow their chosen path. Finally, a feed-forward control strategy is shown to reduce the load transfer across the rear axle by an additional 30% in transient situations, but only if the steer input signal is sufficiently free of noise.

Carbody tilting is today a mature and inexpensive technology that allows higher train speeds in horizontal curves, thus shortening travel time. This doctoral thesis considers several subjects important for improving the competitiveness of tilting trains compared to non-tilting ones. A technology review is provided as an introduction to tilting trains and the thesis then focuses on enhancing the benefits and strategies for less motion sickness. A tilting train may run about 15% faster in curves than a non-tilting one but the corresponding simulated running time benefit on two Swedish lines is about 10%. The main reason for the difference is that speeds are set on other grounds than cant deficiency at straight track, stations, bridges, etc. The possibility to further enhance tilting trains’ running speed is studied under identified speed limitations due to vehicle-track interaction such as crosswind requirements at high speed curving. About 9% running time may be gained on the Stockholm–Gothenburg (457 km) mainline in Sweden if cant deficiency, top speed, and tractive performance are improved compared with existing tilting trains. Non-tilting high-speed trains are not an option on this line due to the large number of 1,000 m curves. Tilting trains run a greater risk of causing motion sickness than non-tilting trains. Roll velocity and vertical acceleration are the two motion components that show the largest increase, but the amplitudes are lower than those used in laboratory tests that caused motion sickness. Scientists have tried to find models that can describe motion sickness based on one or more motion quantities. The vertical acceleration model shows the highest correlation to motion sickness on trains with active tilt. However, vertical acceleration has a strong correlation to several other motions, which precludes vertical acceleration being pointed out as the principal cause of motion sickness in tilting trains. Further enhanced speeds tend to increase carbody motions even more, which may result in a higher risk of motion sickness. However, means to counteract the increased risk of motion sickness are identified in the present work that can be combined for best effect. Improved tilt control can prevent unnecessary fluctuations in motion sickness related quantities perceived by the passengers. The improved tilt control can also manage the new proposed tilt algorithms for less risk of motion sickness, which constitute one of the main achievements in the present study. Local speed restrictions are another means of avoiding increased peak levels of motion sickness when increasing the overall speed. The improved tilt control and the proposed tilt algorithms have proven to be effective in on-track tests involving more than 100 test subjects. The new tilt algorithms gave carbody motions closer to non-tilting trains. Rather unexpectedly, however, the test case with the largest decrease in tilt gave a greater risk of motion sickness than the two test cases with less reduction in tilt. It is likely that even better results can be achieved by further optimization of the tilt algorithms; the non-linear relation between motions and motion sickness is of particular interest for further study. / QC 20110429

In the modern day, recognition and tracking of face or the iris is potentially one of the most powerful ways of differentiating between an authentic person and an imposter. Our method uses stereo vision to track the 3-Dimensional coordinates of a target equivalent to a person’s eyes and using a pan-tilt unit we target these areas for additional processing such as iris or facial imaging. One of the most important parts involved in tracking is the way the pan-tilt unit is calibrated. There have been techniques in the past where PTZ (Pan-tilt-zoom) digital camera has been used and calibrated using self calibration techniques involving a checker board calibration grid but the tracking error was found to be large in these techniques. We introduce a more accurate form of calibration of the pantilt unit using photogrammetric calibration technique and view the pan-tilt unit as an emulation of a Pinhole Lens Model to detect and track the target. The system is demonstrated on ideal targets.

Syftet med denna studie är att undersöka skillnader mellan en geodetisk och en icke-geodetisk metod för deformationsövervakning. För att undersöka detta konstruerades två experiment som båda efterliknar deformationsförloppet för en lutande vägg. Experiment 1 simulerades med hjälp av en whiteboardtavla som roterades runt en axel, där två sensorer placerades längst med rotationsaxeln. I experiment 2 placerades tre sensorer högst upp på en plywoodvägg, som sedan lutades med passbitar. Tanken var att simuleringarna skulle efterlikna en lutande vägg som deformeras över tiden. Alla mätningarna gjordes vid två olika tidpunkter eller epoker. Den geodetiska metoden bestod av att mäta längder och vinklar med en totalstation, vilka sedan användes för att beräkna väggens lutning. Inmätningen skedde via rutnätskanning och en minstakvadratanpassning av ett plan, från vilket trigonometriska sedan funktioner användes för att beräkna väggens lutning mot zenitriktningen, vilken benämns β. Den icke-geodetiska metoden utgjordes av sensorsystemet FlatMesh, som består av tre lutningssensorer samt en 3G-gateway. Utöver detta tillhandahålls en online-monitor som sammanställer och presenterar sensorernas data. Förhållandet mellan horisontalplanet och sensorns Y-axel bedömdes motsvara totalstationens vinkel β. Experiment 1 visade på en skillnad mellan de olika metoderna på enbart 13,392 bågsekunder, vilket motsvarar 0,065 mm/m. Experiment 2 visade på en skillnad om 316,08 bågsekunder eller 1,5324 mm/m. Detta visar att sensorerna ger olika resultat beroende på hur de placeras på objektet, ur vilket slutsatsen kan dras att sensorerna inte lämpar sig för monitorering av objekt där riktningen på deformationen ska studeras. Däremot lämpar de sig för att upptäcka mycket små lutningsförändringar på objekt. / This study aims to determine the differences between a geodetic and a non-geodetic method for deformation surveying. Two experiments were performed to investigate whether the locations of the sensors on the deformation object influence the results. The first experiment used two sensors that were mounted at the object’s rotation axis. The second experiment were performed using a plywood wall that could be tilted. In this experiment, three sensors were placed on the upper parts of the board. All measurements were carried out over two epochs. The geodetic method consisted of using a total station followed by a least-squares adjustment to calculate the tilt of the wall (denoted by β). The non-geodetic method was to use the sensor system FlatMesh. This system consists of three tilt sensors and a 3G-gateway. In addition, a web monitor is included, which enables computation and presentation of the sensor data. The tilt that was determined to best resemble the total station tilt β were the tilt of the sensor’s Y-axis in relation to the horizontal plane. According to the two surveys, the following can be stated. The first experiment showed a difference between the methods of 13.392 arcseconds, which corresponds to 0.065 millimetres per meter. The second experiment resulted in a difference of 316.08 arcseconds or 1.5324 millimetres per meter. This indicates that the sensors give different results depending on how or where they are mounted. It is concluded that the sensors are not suitable for monitoring objects where the direction of the deformation is of interest. However, they are suitable for detecting very small slope changes on objects.