In manual control, the welding gun’s moving speed can significantly influence the welding results and critical welding operations usually require welders to concentrate consistently in order to react rapidly and accurately. However, human welders always have some habitual action which can have some subtle influence the welding process. It takes countless hours to train an experienced human welder. Using vision and IMU sensor will be able to set up a system and allow the worker got more accurate visual feedback like an experienced worker. The problem is that monitor and measuring of the control process not always easy under a complex working environment like welding. In this thesis, a new method is developed that use two different methods to compensate each other to obtain accurate monitoring results. Vision sensor and IMU sensor both developed to obtain the accurate data from the control process in real-time but don’t influence other. Although both vision and IMU sensor has their own limits, they also have their own advantage which can contribute to the measuring system.

<p>Rotundus AB är ett världsledande företag inom området för sfäriska robotar som håller på att utveckla den markbundna farkosten GroundBot(tm). För att detektera robotens translations- och rotationsrörelser används accelerometrar och hastighetsgyron. Dessutom används sensorerna för tröghetsnavigering när GPS-täckning saknas.</p><p>Målet med projektet innebär att ta fram en metod för att utvärdera precisionen hos sensorer för robotens tröghetsnavigeringsenhet. Metoden koncentrerar sig på fyra av sensorernas felkällor: Skalfaktor, bias, slumpvandring och biasinstabilitet.</p><p>För att beräkna skalfaktorn och biasen används linjära minsta kvadratmetoden där utsignalen från sensorn anpassas mot ett teoretiskt värde. Slumpvandringen och biasinstabiliteten utvärderas med Allans varians.</p>

Rotundus AB är ett världsledande företag inom området för sfäriska robotar som håller på att utveckla den markbundna farkosten GroundBot(tm). För att detektera robotens translations- och rotationsrörelser används accelerometrar och hastighetsgyron. Dessutom används sensorerna för tröghetsnavigering när GPS-täckning saknas. Målet med projektet innebär att ta fram en metod för att utvärdera precisionen hos sensorer för robotens tröghetsnavigeringsenhet. Metoden koncentrerar sig på fyra av sensorernas felkällor: Skalfaktor, bias, slumpvandring och biasinstabilitet. För att beräkna skalfaktorn och biasen används linjära minsta kvadratmetoden där utsignalen från sensorn anpassas mot ett teoretiskt värde. Slumpvandringen och biasinstabiliteten utvärderas med Allans varians.

Humanity has been fishing for a long time, more specifically underice with nets, this could be a lot more efficient if it could beautomated in any stage of the procedure. The project aim is toconstruct an automated system for assisting in net fishing under ice.A navigation system using a GPS module, an IMU and radio transceiverwas designed, constructed and implemented on a car for testing. Theresult was a system that could navigate to either a predeterminedpoint or return to a point saved by the system. A "manual mode" wasalso implemented to be able to operate the system as a remotecontrolled car with a computer coupled to an Arduino with a radiotransceiver sending control signals to the prototype. The prototypeworked as intended, but would need further improvements to theautomated navigation system's precision to be ready for commercialuse. This should be the primary focus of any further development ofthis project.

The primary purpose of this thesis work is to design and implement a system supporting the fire-fighter’straining evaluation. The solution presented in this report comprises a real-time locating system andInertial Measurement Unit that collect the training data and an application built with the Unity3D gameengine used for processing. Before the presented application can process any data, a training facility must be recreated, and atrainer or an expert must define training guidelines, such as zones that a trainee must visit, objectshe/she has to see, and the required execution time. After this preparation and completed trainingsession, collected data is processed and the visualization together with evaluation process begins. Theprogram visualizes trainee’s position and his/her field of view and evaluates his performance. For thelatter, it uses the Unity3D collision detection feature. After the evaluation, the trainer can see detailedresults with recorded time stamps, and using them, he/she can fast forward to specific, important points,reducing the time to analyze trainee’s performance throughout the whole training. Tests which the authors conducted confirm that the built system meets requirements and is useful forevaluation of the trainee’s performance.

Understanding the behavior of pedestrians in road transportation is critical to maintain a safe en- vironment. Accidents on road transportation are one of the most common causes of death today. As autonomous vehicles start to become a standard in our society, safety on road transportation becomes increasingly important. Road transportation is a complex system with a lot of dierent factors. Identifying risky behaviors and preventing accidents from occurring requires better under- standing of the behaviors of the dierent persons involved. In this thesis the activities and behavior of a pedestrian is analyzed. Using sensor data from phones, eight dierent events of a pedestrian are classied using machine learning algorithms. Features extracted from phone sensors that can be used to model dierent pedestrian activities are identied. Current state of the art literature is researched to nd relevant machine learning algorithms for a classication model. Two models are implemented using two dierent machine learning algorithms: Articial Neural Network and Hid- den Markov Model. Two dierent experiments are conducted where phone sensor data is collected and classied using the models, achieving a classication accuracy of up to 93%.

This thesis is a feasibility study of contemporary indoor positioning approaches in an hospital environment using sensor available on Android phones together with Wi-Fi fingerprintingand map information. The purpose is to determine the resolution of pedestrian indoor positioning and whether it is sufficient for room level accuracy. Accurate and robust positioning for outdoor applications based on mobile networks and satellite systems, such as the Global Positioning Service (GPS), has been around for many years. However these systems are not suitable for positioning inside buildings due to a high level of signal degradation. Through the years various pedestrian indoor positioning methods have been proposed.A simple algorithm for suppressing random movement of the mobile phone is tested. Two versions of the Extended Kalman Filter (EKF) are compared for fusing the Inertial Navigation System (INS) measurements during Pedestrian Dead Reckoning (PDR). The TRIAD algorithm is tested for suppressing the effects of large magnetic disturbances. Wi-Fi fingerprinting using two combinations of positioning algorithms and radio maps is tested. The EKF is tested for fusing PDR and Wi-Fi fingerprint position estimations. The Particle Filter (PF) is tested for combining PDR with Wi-Fi fingerprint positioning with a geometrical map. Static Received Signal Strength Indication (RSSI) measurements are carried out to detect variable Wi-Fi transmission power. The results show that adding more informations sources improves the positioning performance. Also fusion using PF outperforms the EKF in more complex indoor environments and movement patterns. / En starkt växande tillgång och kapacitet hos trådlösa nätverk i kombination med explosionen inom mobiltelefoni, i synnerhet vad gäller smartphones, har lett till ett enormt ökat intresse för och utveckling inom området inomhuspositionering. Det har under många år funnits lösningar för positionering i utomhusmiljöer, exempelvis GPS och triangulering med mobila basstationer, men inga av dessa system lämpar sig för inomhuspositionering eftersom signalerna tappar alldeles för mycket i intensitet när användaren befinner sig inomhus. Under årens lopp har flera olika lösningar för inomhuspositionering föreslagits. I denna uppsats testas olika lösningar för inomhuspositionering med smartphones i en sjukhusmiljö. Testen baserar sig på de sensorer som finns i en smartphone med operativsystemet Android i kombination med Wi-Fi triangulering och en digital planlösning över testområdet. Syftet är att undersöka om noggrannheten kan bli såpass så bra att en upplösning på rumsnivå uppnås. En enkel algoritm för att kompensera för slumpartade och oplanerade rörelser hos mobiltelefonen testas.  Två versioner av det utökade Kalmanfiltret testas för tröghetsnavigering. TRIAD algoritmen testas för att motverka magnetiska störningar. Två kombinationer av radiokartor och positioneringsalgoritmer provas för att genomföra Wi-Fi positionering. Ett utökat Kalmanfilter används för att kombinera resultaten av tröghetsnavigeringen med Wi-Fi positioneringen. Ett partikelfilter används för att utföra sensorfusionen av tröghetsnavigeringen, Wi-Fi positioneringen och den digitala planlösningen. Resultaten visar att ju mer information som tillförs under positioneringen desto större blir noggrannheten samt att partikelfiltret ger en bättre noggrannhet i en komplex inomhusmiljö i kombination med komplicerade rörelsemönster än det utökade Kalmanfiltret.

The diploma thesis deals with possibilities of tracing the trajectory of mobile mapping systems in order to compare their accuracy. This thesis discusses the individual components of the mobile mapping device that were used in its construction. The thesis contains the results of the experiments that were performed on individual test.

This document describes a device capable to determine geographical position thanks to its GNSS module and measure the change of movement with the help of an accelerometer and gyroscope. Outputs of these integrated circuits are combined, data are processed and fused thanks to a numerical integration and mathematical filters. Results are stored on a memory card. The whole development is described from a concept, through the making of the device and software algorithms to the testing of its functions.

This paper deals with the application of SLAM algorithms on experimental four wheel vehicle Car4. The first part shows the basic functioning of SLAM including a description of the extended Kalman filter, which is one of its main components. Then there is a brief list of software tools available to solve this problem in the environment of MATLAB and an overview of sensors used in this work. The second part presents methodology and results of the testing of individual sensors and their combinations to calculate odometry and scan the surrounding space. It also shows the process of applying SLAM algorithms on Car4 vehicle using the selected sensors and the results of testing of the entire system in practice.