Pursue

Pursue is a development framework for service robotic applications in which robotic agents interact with humans and both collaborate to achieve common goals, such as “the robot should carry an item without bumping into the humans moving in the ward”. Pursue applications can be designed effectively using Timed Games and controlled at runtime through game strategies that govern the collaborative actions of the agents. The development of Pursue applications follows a two-steps approach whereby the application is first specified by a domain specific language (Pursue-ML) and analyzed for feasibility; and secondly the application is automatically deployed in a realistic scenario (the target robotic platform is TurtleBot).
If the first phase is successful, the application controller, which allows the application to fulfil the desired goal, is created. The deployment phase is realized by automatically translating the controller into executable ROS code and by setting up the ROS nodes that manage the application at runtime.

Possible thesis are related to the following activities:

• improving Pursue-ML to capture more complex forms of collaborations and scenarios;
• improving the translation of game strategies into executable ROS code;
• improving the automatic deployment to create digital-twin based applications in which physical agents operate in the real environment while virtual ones operate in the CoppeliaSIM virtual scene.

Reference: https://doi.org/10.1007/s00165-020-00509-0
Related: https://doi.org/10.1016/j.robot.2023.104387
Skills: Automata Theory and Formal Modeling; Planning/Optimization Tools; Basic Programming Skills (the specific language may vary depending on the chosen technologies)

Est. Effort: 10(+/-)2 months (full time)