Subsea petroleum extraction systems may be large and complex, and many decisions affect the production. Maintaining high production levels is not a trivial task. As decisions are made based on available information and experience, better decisions come with better information. Decision support tools may provide essential information to achieve better production levels.In this master thesis, different methods are proposed as decision support tools. The aim is to increase the production from a part of a subsea production system, consisting of a manifold with seven producing wells and two flowlines, given certain system constraints. The methods are based on well models and numerical optimization, and both static and dynamic optimization is considered. The well models are non-linear, and binary decisions are also present. The problems that arise are complex MINLP problems, and are solved by combining ’brute force’, ’Branch & Bound’, and a nonlinear solver. The solution of the problems is implemented in MATLAB, and tested on predefined test scenarios, with no, little or extensive dynamics present. The performance is assessed by simulations, and by calculating the resulting average production.It was found that static optimization to decide the well settings, such as valve openings and flowline routing, has a great potential to increase the oil production from the system. The results when applying a dynamic approach to the system were not conclusive, but the methods proposed showed no indications of any major performance increase, relative to applying only static optimization.

This report studies the development of a Multiplicative Extended Kalman Filter for orbit and attitude estimation for the 10times10times20 cm CubeSat at the Norwegian University of Science and Technology (NTNU). The filter was developed in a tightly coupled manner with respect to the GPS attitude solution, based on data from differential carrier phase measurements. These measurements are aided by measurements from a three-axis magnetometer, and inertial measurements from a gyroscope. Four antennas are virtually mounted on the satellite to obtain three baselines of 1 m each. The MEKF is complemented by an integer ambiguity resolution method, which makes sure that the solution for a GPS signal is not accepted until the integrity check value for all baselines is below the acceptance threshold. Until the ambiguities are resolved, the Multiplicative EKF is reliant upon the gyro measurements, and the magnetometer. The filter has been simulated with various attitude maneuvers.The MEKF performs orbit estimation based on measurements from GPS position, velocity, and timing data, from which it estimates the Keplerian orbital parameters to determine the orbit of the craft. It operates as an ordinary EKF for this purpose. Simulation shows that the filter is able to determine the attitude and orbit of the spacecraft from the given measurements, and that it is robust to a temporary loss of the GPS measurements. However, the orbit estimator assumes a circular orbit. The quality of orbit estimates are therefore dependent on the eccentricity of the orbit.

Eurobot is an annual competition for autonomous robots. Typically two teams compete against each other for 90 seconds on a 2x3m playing area. The main goal is to collect as many points as possible. There are several matches to determine which robot is the best. The rules are different every year. An autonomous robot for this year's competition was designed and built. The rules for Eurobot 2012 were studied and a design concept was created. In order to implement the design a series of technical pieces of work was carried out. The tasks involved several fields of study including engineering cybernetics, electrical engineering, computer science and mechanical engineering. A laser-tower positioning system from 2010 was further developed and improved. In addition the robots drive wheels hall-sensors were used to compensate for the robots movement. An extended Kalman filter was created to transform these measurements into a position and orientation estimate. Two PID regulators were used to maneuver the robot, one regulator controlling the rotation, the other translation. A circuit board following the EPIC-plus standard with several efficient power supplies, a micro controller and a circuit board stack was designed, produced and tested. This circuit board was driven by a lithium battery and acted as a power supply for-, and took care of low level interaction with, all motors, servos and actuators on the robot. Firmware was implemented on the circuit board that provided an interface to control all hardware via CAN-bus to a tablet PC. To implement strategic choices, algorithms and artificial intelligence, an elaborate software system was created. The high-level programing was done in Go, a new and exciting programing language from Google. A featured Debug-GUI that presented real-time information and allowed for robot interaction was provided. A strong focus on design and a test driven development, resulted in robust and stable software. A mechanical design of the robot was created in collaboration with a group of students through the course TTK4850 ``Experts in team'' at NTNU. This work lead to a complete robot with a clean implementation hosting advanced technical solutions. The final software allowed strategies to be reprogrammed before each match, and the physical robot was easily maintainable. The positioning system can move the robot to any coordinate on the playing area. In the Eurobot 2012 competition, the robot won three out of five matches and ended at 23rd place out of 43 international teams.