Skip to main content

Towards Intelligent Cognitive AUVs

Periodic Reporting for period 2 - TIC-AUV (Towards Intelligent Cognitive AUVs)

Reporting period: 2018-03-01 to 2019-02-28

Recent advances in robotics have resulted in machines capable of performing many tasks autonomously. However, a significant obstacle to intelligent robots being used in real-world scenarios is their limited ability to cope with unexpected events and environments, to deal with faults, and to make smart decisions in response to changes in the world. The aim of this project is to develop a semantic knowledge-representation system that will make it easier for robots to handle such situations, and therefore achieve more persistent and long-term autonomy. An ontological representation for semantic data is used for several reasons:

- it is easy to create and examine the concepts used by the system, and the attributes available for each concept, being a standard language with clear rules and widely adopted;
- many tools exist to perform logical reasoning within the knowledge base;
- an ontology represents a well-specified central data store that all the software components comprising the agent can make use of;
- common ontologies are very useful for exchanging data between robotic systems created by different organisations.

A proper knowledge representation system is pivotal for long term autonomy as it allows to store information, to reason on the acquired information and to augment the knowledge through reasoning. This further allows to have smarter planning and control systems which can query the knowledge base and be notified about important information relevant to the current plan.

Overall, the concrete objectives of the project are:

- develop a probabilistic world model system, which addresses both the external world, in order to
aid localisation and to dynamically check consistency of current plan; and on the internal state of the
vehicle, in order to aid internal fault management;
- develop an active localisation system, based on both geometric and semantic information, which
is general enough to accommodate for various and heterogeneous types of actions;
- develop a fault management system, based on a probabilistic world model and –
where applicable – on localisation information, in order
to address a wide range of faults, from single component up to task level failures.

The project has met the objectives, with encouraging results in semantic localisation and in fault management. Additionally, it has expanded its original scope to address localisation of Autonomous Surface Vessels in GPS-denied environments.
The main results of the projects are:
- development of a semantic world model
- development of a series of algorithms for localisation of marine robots, including underwater robots (AUVs) and surface vessels (ASVs)
- development of an integrated framework for fault management
- integration and test with real robots in the field and in water tanks.

The project content and results were widely disseminated, ranging from scientific conferences and publications, to specialised workshops, to ad-hoc meetings and presentations.
The highlight in the dissemination activities has been the presentation at the United Nations in New York, at the General Assembly during the UN Oceans Conference, linking development of advanced marine robotics with the fulfilment of the UN Agenda 2030 on sustainable development, with a focus on SDG14 Oceans.
"Robot localization has been researched by scientists from the very beginning of robotics. Knowing its position and orientation is an essential task for an autonomous system in many circumstances.
Very often, it is a prerequisite to perform more complex tasks.
The marine domain is no exception. In most - if not all - real life scenarios, the robot location is an essential information.
Marine robots are increasingly used to perform a great variety of tasks, ranging from oil\&gas applications to defense, from marine biology to underwater archaeology. In all of these scenarios, the robot location is fundamental.
There are several challenges to perform underwater localization. The lack of GPS signal is the most evident one. In order to overcome this, various acoustic-based solution can be employed, like for example Long BaseLine (LBL) acoustic positioning system. This requires to deploy acoustic transponders as aid for the vehicle, which can compute its location with a triangulation from the data received by the transponders. The drawback of this technique is however the need to actively deploy external transponders, which cause additional cost, time and logistic challenges. Additionally, the GPS location at the drop-in point might not be the same than the GPS on the seabed, especially in deep sea, with strong currents.
Many offshore infrastructures are located in environments which fall into this category.
For this reason, several techniques have been used in the past years to allow an underwater vehicle to determine its location based entirely on the on-board sensor suite.
Geometric approaches were developed based on distance sensors like sonar, in parallel with geometric approaches developed in land robotics, based on laser scanners.
In recent years there has been a substantial interest from the research community to explore semantic aspects of knowledge representation, and its influence in the vehicle's tasks. Generally speaking, robots still lack the high-level abstraction capability typical of humans. This is a complex problem, as it aims to shift the paradigm from sensor processing into a more organized, long-term knowledge structure in robotics systems, with possibility of augment, reasoning and learn.
This project represents a step in this direction, with the of use of semantic information in processes traditionally covered only by geometric approaches.
To the best of the researcher's knowledge, no significant semantic approach for localization in the marine robotics domain is currently available in the related literature.

Several application domains can benefit from the research carried out in this project. Inspection of subsea structures such as oil platforms using autonomous underwater vehicles (AUVs) is a prominent example. In this domain, the original plans for structures can generally be made available to the AUV in advance of a mission. However, frequently these structures have changed or moved when the AUV actually reaches them. If the robot has a pre-programmed survey path, unexpected obstacles can cause it to abort the mission. Reactive obstacle avoidance algorithms can improve the robustness but a proper knowledge representation framework, able to incorporate newly discovered information, is desirable when the aim is for the robot to be ``intelligently"" autonomous, situation-aware and able to use dynamic world knowledge. Moreover, fault awareness - let alone fault management strategies - has been explored in research but very rarely integrated in a more complex system.
Other important application areas in the marine domain which would benefit from this research are defence, archaeology and marine biology. In all these domains, a cognitive vehicle would export capabilities and functionalities which are currently out of the reach of conventional commercial AUVs, unlocking greater opportunities in the field, reducing risks and cost.

On a societal dimension, this work is very well aligned with the United Nations Sustainable Development Goals (SDG), in particular with SDG14 - Oceans, and was even featured at the UN Oceans Conference."