The main objective of this thesis was to combine the theory on synchronization of robot manipulators with the concept of hybrid force/motion control; resulting in a controller capable of following both the trajectory of another robot and a desired force trajectory at the same time. This report includes a short introduction to synchronization theory for robot manipulators, and a more thorough summary of existing hybrid control schemes. An intuitive method for describing constraints caused by the environment is presented, and this leads to a straightforward way to separate the interaction control problem into two subproblems. When the manipulator end-effector is in contact with a surface, the hybrid controller tracks a position reference along the surface and a force reference normal to the surface. The contribution of this thesis is a proposed hybrid controller, incorporating ideas from previous hybrid schemes. The closed loop system is shown to be globally asymptotically stable in the position controlled degrees of freedom, and bounded in the force controlled degrees of freedom. The performance of the proposed hybrid control concept is demonstrated by simulations. The proposed hybrid controller easily reduces to a pure position controller when desired.

This thesis addresses the problem of robot synchronization with obstacle avoidance. While these two fields have been studied extensively on their own, they have not yet been considered together as one problem. This thesis is divided roughly into four parts which are to some extent self contained. The theory is presented in a narrative that culminates with the stability proof of the proposed controller. Examples and figures are used in order to keep the material manageable and readable. The introductory part of the thesis consists of chapters 1 and 2. We present the notation and some mathematical background which is necessary for the theoretical analysis. We go on to review the diversity of ways in which one may approach this problem from a control design standpoint. We derive the robot dynamical model in chapter 3 as well as solve other modeling specific problems. This chapter is of little theoretical interest, but is needed to implement a simulator on which we may test our controller. This chapter contains no new contributions but can be read as a guide to robot modeling. The first contributions in this thesis are found in chapter 4 where we propose a real time implementable solution for solving the shortest distance estimation problem. It is important to know the distance to an obstacle in order to avoid it. The solution is a dynamic implementation of a steepest descent optimization scheme which is suitable to run on-line. Chapter 5 is an introduction to the involved control design found in chapter 6. We review results from obstacle avoidance literature and argue for our choice of using the task space control design method. The main contribution of this thesis is found in chapter 6. A controller is developed and is shown to produce a stable closed loop system. We first develop a controller considering only collision for the end effector, and then we extend this to work with full robot collision. The response of the robot is such that it will track a reference trajectory whenever it is locally possible. When one cannot track the reference trajectory because of obstacles hindering the movement, then the trajectory is tracked in all directions in which the robot can move freely. The controller is simple and elegant, and does not rely on heuristics common in traditional solutions to obstacle avoidance control.

The main purpose of this report, is to evaluate feasibility of using The Global Positioning System as an aiding tool for vehicle state estimation based on nonlinear techniques, and to develop a potential solution to the road bank angle problem. Previous work within the CEmACS project, includes development of a general nonlinear observer for lateral and longitudinal velocity, and an augmentation in the form of road-tire friction adaptation. Because the existing solutions have been shown to lack robustness with regards to certain disturbances, such as road grade and road bank angle, it has been stated that the estimation scheme should be upgraded, so that these disturbances can be accounted for. By including GPS velocity or a or a roll rate gyro measurement in the observer feedback loop, the possibility of detecting previously inobservable quantities is gained. In simple terms, evaluation of feasibility corresponds to demonstrating improvements and limitations of new solutions, using relatively crude methods in the test procedures. Problems related to the above mentioned task, are approached by means of signal processing and control theory. Following an intuitive sequence of operations, the report presents GPS theory and results first, as this lays the foundation for all subsequent results. Methods used comprise simple differentiaton, rotational kinematics and discrete filtering. Secondly, theory and results related to nonlinear observers, with focus on GPS aiding, are examined thoroughly. Lyapunov theory, known from control engineering, is used to evaluate stability, while data from simulations and actual vehicle tests is used to show how a new observer scheme can improve existing solutions. Before the most important results are presented, something should be said about their accuracy and significance. It has already been pointed out that the methods used are not based on optimality requirements, and consequently the results are best viewed as indicators of potential, rather than absolute solutions. This is especially true for the GPS velocity calculations, which are based on differentiation of position measurements; generally not a desired approach. In this report, it is firstly shown that GPS position measurements can be used to compute receiver velocity in the body-fixed coordinate frame. While this is a crude approach, resulting in relatively poor signal to noise ratio, it is easily implemented on low-level equipment. It is also shown that it is possible to use these velocities as measurements in a nonlinear observer structure, slightly modified from previous solutions within the CEmACS project. By doing this, accurate estimates of road grade and bank v angle are achieved, so long as these vary slowly enough. Stability of the observer is not proven in the general sense, but it is shown that it can be made stable through realistic assumptions and gain selection. Stability is further demonstrated through the use of data sets from actual vehicle tests. Secondly, a mathematical model of roll dynamics is combined with a roll rate measurement to create the possibility of detecting road bank angle. This is done by the now familiar nonlinear observer approach. Usefulness is demonstrated by simulations, but no stability proof is presented. The main conclusion is that it is feasible to use GPS aiding to account for robustness problems in a vehicle state estimation scheme, and that the inclusion of a roll rate measurement opens up possibilities for “cheap” bank angle detection. A direct implementation of results presented in the report may not be ideal, but the fact that the system works for a large set of conditions, suggests that it is worthwhile to develop it further. This is especially true when it is assumed that GPS receivers will become an integral part of new vehicles in the near future. Refinements and upgrades can be made in the form of more advanced GPS technology, new parameter estimation techniques and integration with the road-tire friction adaptation scheme.

This thesis deals with the mathematical modeling of hydraulic heave compensation systems. When performing operations such as launch and recovery of remote operated vehicles and lowering subsea installation parts to the sea floor, it is important to attenuate unwanted load motion caused by elongation of the cable and heave motion of the vessel. Quite often, such operations must be put off while waiting for the weather to calm down. Extending the window of operations by developing equipment that can handle varying sea states can result in significant cost savings. There exist both electric and hydraulic heave compensation devices, but the main focus of this thesis is on the hydraulic configurations. A mathematical model is developed in Simulink, and simulations are performed for long-crested seas with values corresponding to the average sea conditions in the North Sea.

An underlying requirement for any Autonomous Underwater Vehicle (AUV) is to navigate through unknown or partly unknown environments while performing certain user specified tasks. The loss of an AUV due to collision is unjustifiable both in terms of cost and replacement time. To prevent such an unfortunate event, one requires a robust and effective Collision Avoidance System (CAS). This paper discusses the collision avoidance problem for the HUGIN AUVs. In the first part, a complete simulator for the HUGIN AUV is implemented in matlab and simulink. This includes a 6 degrees-of-freedom nonlinear AUV model, simulated environment including bottom profile and surface ice, navigation- and guidance functionality and sensor simulators. In the second part a number of well known strategies for the collision avoidance problem is presented with a short analysis of their properties. On the basis of the implemented simulator, a proposed CAS is developed and it’s performance is analyzed. This system is based on simple principles and known collision avoidance strategies, in order to provide effective and robust performance. The proposed system provides feasible solutions during all simulations and the collision avoidance maneuvers are performed in accordance with the specified user demands. The developed simulator and collision avoidance system is expected to provide a suitable framework for further development and possibly a physical implementation on the HUGIN AUVs.

Design of a heading autopilot and way-point navigation system an underactuated USV.

Pipelines with multiphase flow will exhibit large and highly nonlinear liquid rates during transients caused by changes in production rate. This requires either a large separator (slug catcher) downstream, capable of handling all disturbances, or, some sort of control of the rate change in order to ensure that the downstream processing equipment can cope with the disturbances. In this report, it is proposed to use smart liquid control to optimally control the production during rate changes. The term smart liquid control was coined to describe the use of nonlinear model predictive control (NMPC) and a nonlinear pipeline model to control liquid levels. The pipeline from the Snøhvit field to the Hammerfest LNG plant was used as a test case on which to implement smart liquid control. The technique was proved using a simplified OLGA pipeline model for both predictions and as process, i.e. perfect model. Good results were shown, especially when compared to manual control. Smart liquid control therefore presents a possibility to reduce separator size for new projects. Another highly interesting use of smart liquid control is to optimize the use of separator buffer capacity in order to maximize production. The NMPC method used in this report, a single shooting multistep quasi-Newton method, was elaborated on in a literature review chapter. Also an implementation of the method was made and described in the context of NMPC. The implementation of the algorithm was made in the in-house STATOIL MPC-tool, SEPTIC. In order to assess the performance of the algorithm, it was tested on a small continuously stirred tank reactor system and the results reported and discussed.

A ship scheduling problem in optimization of water-borne crude oil transportation has been investigated. The classic optimization problem the most closely related to the problem at hand is the Multi-Vehicle-Pick-up-and-Delivery Problem with Time Windows (m-PDPTW). In addition to the basic characteristics of the m-PDPTW, the studied problem has an additional degree of freedom due to having pick-ups and deliveries that are not matched. This extra freedom gives new possibilities when creating effective heuristics when dealing with transportation problems. The studied problem has been presented in relation to carefully selected background literature. On this basis a proposed heuristic has been developed, and implemented using some already existing structures in the commercial decision support system TurboRouter. The studied problem is an industrial shipping problem, an operational mode where the shipper owns the cargo to be transported. No income is therefore made directly from transporting goods. Therefore the objective function chosen was net income, which in this mode is the same as minimizing the net expenses. A multi-start local search with pre-matching of pick-ups and deliveries heuristic was chosen based on an assessment of problem size, problem type, real life applicability and existing software. This heuristic consists of three main parts. First the pick-ups and deliveries are matched and merged in a pre-matching heuristic, and then a large number of initial solutions are generated by an insertion heuristic. The best initial solutions are then improved by a local search. Two strategies were developed for pre-matching and then tested. The one with the best test results was subsequently used in the heuristic. This multi-start local search with pre-matching of pick-ups and deliveries heuristic has been subject to rigorous testing and was compared to a single-start local search and multiple initial solutions heuristic. The solutions generated by the multi-start local search heuristic were superior compared to those of the other heuristics, but the computation time necessary was high and higher than those of the heuristics which it was compared to. This high computation time is partially believed to be a result of flexible data sets resulting in broad solution spaces. In addition some computationally expensive heuristics were deployed, increasing the computation time. In real life applications, finding a solution relatively quickly is of importance. Therefore the heuristic may need to be simplified and used on "tighter" data sets than some sets used in testing to be real life applicable.

In offshore production systems for oil and gas, the wells are usually controlled manually, while the topside process is controlled using a decentralized control system. It is clear that this control structure may not be economically optimal. In the work of this thesis, one decentralized and two MPC based control systems were designed, integrating the control of the production wells and the topside process of an offshore production system. The performance of the controllers was tested in a simulation study with three different disturbance scenarios, defined in collaboration with Hydro. The simulations were also carried out using a manual control scheme, enabling a performance comparison between the designed controllers and the control conventions of today. The automatic controllers showed a significant increase in oil throughput over the manual control scheme, while performing quite similarly when compared to each other. It is however expected that the MPCs will outperform the decentralized control system in a case with a larger number of wells than was used in the simulations of this thesis. The potential of achieving additionally smoothened topside outflow rates by applying a supervisory MPC for topside setpoint manipulation was also demonstrated.

Handling slug flow is a matter of necessity in the oil and gas industry. Large slugs cause problems at the production platform by overfilling the first stage separator leading to poor separation and pressure increase, which again could lead to more flaring or even complete process shut-down. Using the riser base pressure as the controlled variable has been shown in several academic papers to have a stabilizing effect on the slug flow. In this thesis it has been concentrated on designing robust anti-slug controllers using this variable as the controlled variable. Experience from Hydro's multiphase rig at their Research Centre in Porsgrunn shows that re-tuning of anti-slug controllers during experiments is necessary to be able to keep the flow stabilized. This is caused by the change in process gain, which is large for operation around small valve openings and low for operation around large valve openings. An OLGA model was developed for this multiphase rig. This model was shown to predict the amplitude of the slugs and their frequencies fairly well, however, the transition point from stable flow to slug flow differed to a larger extent. Several gain scheduling controllers were designed and tested on this model. The best gain scheduling approach which was found was based on a gain scheduling regime controlling the riser base pressure in a secondary loop without integral action, with a slow primary loop controlling the valve opening (with integral action). The scheduling variable was a low-pass filtered measurement of the valve opening. This approach was also robust with respect to noise and lowered inlet rates.