Research

Project manager: Kristin Y. Pettersen

To achieve truly autonomous resident robots at subsea oil and gas factories, the problems of autonomous docking and autonomous intervention need to be solved. A permanent docking station on the seafloor, where the vehicle can charge the batteries and transfer the results of a mission, will reduce the current need for frequent launch and recovery operations of ROVs and AUVs. For fully autonomous resident robots, autonomous docking is required. The lack of precise positioning systems such as GPS underwater, represents one of the most challenging aspects of a docking operation. This can be compensated for by using alternative positioning methods, a summary of which can be found in [PSSL14]. Close to subsea templates, acoustic signals may be distorted, and thus vision-based methods are preferable. In this work package we will develop accurate and robust methods for vision-based autonomous docking.

Moreover, autonomous intervention is required for truly autonomous resident robots, while existing resident subsea robots are still remotely operated. Interaction tasks such as connector plugging/unplugging and valve-turning are demonstrated experimentally for an ROV in [YRPS17] where they are solved as position control of the end-effector or its wrist-joint, respectively. A lighter underwater robot would, however, experience significant reaction forces which would move its base at the expense of moving the valve. Moreover, no forces and torques are considered in [YRPS17]. In this work package we will target the research question of how to achieve high-accuracy position and force control for autonomous intervention operations with light ROVs and AIAUVs.

Project manager: Kristin Y. Pettersen

In this work package we will investigate cooperation between heterogeneous teams of underwater robots, e.g. consisting of ROVs and AIAUVs, operating together to perform subsea tasks. Cooperation between the ROV and the AIAUV can be advantageous for both inspection and intervention tasks and will provide capabilities that exceed those of any existing marine robots. For instance, the AIAUV can provide the ROV an extra eye, keeping the object and ROV manipulator arm within field of view at all time, giving better situational awareness; The ROV manipulator and the AIAUV manipulator can cooperate, efficiently transporting objects together; Or the AIAUV manipulator can access areas inaccessible to the ROV, while the ROV provides situation overview.

Cooperative operations require tight synchronization between the motion of the ROV and the AIAUV, utilizing recent advances in cooperative control of nonlinear dynamical systems. Existing methods for cooperative control of robot manipulators mainly concern similar and fixed-base manipulators. In this work package, these methods will be extended to provide cooperative control between heterogeneous teams of underwater robots.

Project manager: Martin Ludvigsen

Adaptive missions and on-board re-planning will be adapted to underwater vehicles performing inspection and maintenance work.

In these missions, operational conditions such as environmental variables like sea state, current and visibility often change. But also, the technical conditions for navigation, communication, energy and diagnostics will change the situation and requirements for completion of tasks. Both internal and external conditions need to be represented in the mission system to be considered planning the operation. In many instances, a hybrid solution combining the deliberate planning layer with a reactive layer handling more urgent issues like alarms and technical faults or threats, will represent an effective solution. The planner architecture that does the deliberate mission planning in this project will be scalable, allowing easy testing and verification. It will be possible to run simple plans and to scale up to tasks that are more complex.

Project manager: Martin Ludvigsen

Controlling and manoeuvring an underwater vehicle would be easy if a complete model of the environment were available. Unfortunately, accurate models of the subsea environment are rare, and the location and orientation estimates of objects and obstacles may be inaccurate. To enable higher levels of autonomy, situational awareness mechanisms need to take into account system and mission internal information, including the scope and sequence of the mission, the vehicle internal condition, energy, communication and navigation status and the status of the environment.

External factors like infrastructure for docking, other vehicles including their status, and objects like pipelines valves, templates or the environment together with the surrounding terrain, need to be processed from measurements to information in the system. In particular, optical and acoustical sensor data will need to be processed to provide information on the current scenery and objects present, determining the status and position of the objects, [KW07].

Project manager: Kjetil Skaugset

The different phases of an AUV mission - departure, docking, transit, station keeping/hoovering, and manipulation - will require different control objectives and control strategies coping with a set of operational, environmental and internal constraints subject to varying elements of risks. Indeed, adding multiple underwater robots (eg. AUVs) as well as mission supporting robots (eg. USVs) on the sea surface and potentially in air (airborne drones) require heightened risk awareness and organization control. Eg. the logistics operation of bringing robotic assets (such as AUVs) to a desired location might call for cooperation between several heterogeneous robots acting in different elements.

Organization of robots to coordinate and collaborate to reach a common mission objective will require new organization layers, optimization, control and computational frameworks. Emerging risks are involved, related to lack of knowledge and operational experience with permanently submerged AUVs and robot organizations, the dependency on complex software-based control systems that may include dynamic learning algorithms, as well as a limited ability to verify the safe performance of such systems.

Crucial for a safe operation is situational awareness for online risk control, where the robot organization should be designed to perform complex tasks under significant uncertainties in the system and when operating in an unstructured environment. The robot organization should be able to handle external events and internal faults including reconfiguration, planning and re-planning, and to learn, adapt and improve. An extra autonomy layer (supervisor) which enables the autonomous control system to online model, estimate and control risk will be developed. E.g., for robot cooperation, transit, safe navigation, energy efficiency and anti-collision are the most important, while for tracking and station keeping, high precision positioning is of highest priority. Avoiding collision in a multi-body dynamic environment where interaction is required for successful mission completion will be of high priority.

Project manager: Asgeir J. Sørensen

AUVs for seabed and water column mapping have been a reality for a while. However, permanently submerged AUVs for inspection and manipulation work are new. Safety and efficiency for such operations require robust software designs that can be updated, tested and verified without bringing the vehicles out of water. Formal methods [MYBA20], e.g. temporal logics and model checking, are used to design and verify properties of control algorithms including finite-state logical behaviours. Even if the design methods may be sound, the implementation in terms of software code and the corresponding configurations may be wrong. In-formal test and verification methods known as e.g. hardware-in-the loop (HIL) testing and software-in-the loop (SIL) testing. HIL and SIL testing have turned out to be efficient test methods of computer-controlled system. However, a challenge may be to test all possible combinations of failure methods in a reasonable time for complex control systems. This WP will address both formal and informal methods for robust design, testing and verification of autonomous control systems of the AUV from the design phase into operation phase that includes upgrades of software.