Team Ephesus

Team Ephesus is the Robocup team of Cognitive Robotics Laboratory at Sabanci University. Our research aspirations are rooted in theoretical AI concepts such as declarative programming, causal logic and ontological reasoning, and we possess expertise on integrating the state of the art automated reasoners for use in robotic domains. Team Ephesus uses an enhanced formal planning, execution and monitoring framework that embeds knowledge representation and automated reasoning in each level of conventional 3-layer planning and monitoring framework, in such a way as to tightly integrate these layers.

Robot competitions focus on the employment of mobile robot systems to perform complex tasks in an unstructured settings. In these competitions, robots are expected to deal with the high complexity and the wide variability of surroundings to perform typical everyday tasks. In particular, these robots are not only required to be able to move through the various regions of their work space without colliding with obstacles or people, but also to robustly complete complex tasks, including but not limited to fetch-and-carry duties, solving challenging puzzles and performing everyday household chores. Moreover, when more than one robotic agents are available in the domain, collaboration of these robots with each other need to be addressed. In all completions, the complexity of the tasks and variability of environment place high demands on the robots’ intelligence and autonomy.

Attending robot competitions expose the details of new robotic domains to our team, provide us an opportunity to investigate the applicability of our approaches to address the challenges in these new domains and to further improve our methods and develop novel techniques. We believe that not only the other researchers may stimulate us to further extend our methods and develop new algorithms, but also our research may inspire other researchers to consider automated reasoners for their systems.

Dynamic Simulation for the Tower of Hanoi Challenge

For the Tower of Hanoi Challenge, we have integrated our hybrid planner with motion and grasp planners of OpenRave and solved the generalized Tower of Hanoi problem, where initial configuration of the disks can be set randomly as long they obey the problem constraints. A sample video of such a dynamic simulation with Kuka youBot mobile manipulator is presented below.

Here is a dynamic simulation for the Tower of Hanoi Challenge that includes navigation with the mobile base.

Physical Implementation of the Tower of Hanoi Challenge

Dynamic Simulation of Fetch and Carry Task for the Robocup@Work Competition

We have extended our planning and execution monitoring framework for the Robocup@Work competition. A sample video of the dynamic simulation of typical fetch and carry tasks in a constrained workspace using Kuka youBot mobile manipulator is also presented below.

Robocup@Work Competition at IEEE/IJR IROS 2012