In this research thesis, we are interested in modeling the interaction between a control program for an underwater robot and its knowledge base. The latter can be seen as a digital twin (a virtual model of the autonomous system and its environment, based on perception) for representing and reasoning about the changing environment of the robot. The topic involves modeling the control program and building a corresponding knowledge base. The capabilities of the solution will be analysed in a simulation environment and using formal methods.
As there are many different use cases for underwater robots there are several scenarios for which the cognitive controller can be designed, e.g. underwater pipeline inspection or efficient surveying of the sea floor. We will define the specific task together with you.
Work on this topic will involve the following:
- Designing the algorithm that the control program will execute.
- Designing the structure of the knowledge base, i.e., identifying the relevant things that need to be represented in this digital twin and the relations between them.
- Implementing the designed controller and the knowledge base and connect them to an existing simulation environment.
- Run test scenarios and investigate how changes in the knowledge base effect the behaviour of the robot.
This thesis is suitable for you if you are interested cognitive robotics control and creating a digital twin knowledge base. Depending on your interest, the project can focus more on simulating the behaviour (requires some interest in using robotic control systems, e.g. [ROS]) or a formal analysis of the system (requires some interest in formal methods).
Background of the project
The topic is realised in the context of the European project [REMARO: Reliable AI for Marine Robotics], which explores how underwater robots can operate safely in places where human presence can be both costly and dangerous, managing on their own without human support.