Katholieke Universiteit Leuven (KUL)

The Katholieke Universiteit Leuven has a research group in robotics since the late 1970s, with a focus on the mechatronics and control-theoretic aspects of force-controlled robots. Since the 1990s, this focus has shifted towards the control, estimation, software engineering and task specification of complex, multi-sensor and multi-task robot applications. Over the last couple of years, the Mechanical and Computer Science (Prof. Luc De Raedt) departments are joining forces to tackle the whole “intelligence stack” of robotics, from realtime reactive control, over on-line intelligent sensor-based skills, to the strategic-level symbolic and logical reasoning.

The expertise of KUL which is most relevant to RoboHow is (i) realtime sensor-based control and (Bayesian) estimation for industrial and (redundant) service robots, (ii) formal specification of single and multi-robot, sensor-based robot tasks, and (iii) software engineering to build complex, multi-component applications. KUL traditionally follows a “bottom-up” approach to unify modelling, specification, estimation, learning, and control of robot tasks, with realtime (open source) implementations for the “lowest” control levels: the reactive level (with direct feedback from sensors to actuators), and the skill level (in which the sensor data are interpreted in the context of a sensori-motor models of a task). The above-mentioned co-operation with the KUL Machine Learning group is tackling the integration between the lower levels and the higher level which is a main focus of the current project. KUL also has a long track record in open source software for robotics, with the integration of the Orocos project with ROS being the most visible and recent one.

KUL will drive integration and harmonization in the project, by collaborating with all partners to consolidate mature motion and action primitives into a generic constraint-based specification and control paradigm. This scientific role is complemented by a technological driving role, as one of the two major software engineering partners in the project, with major focus on: (i) avoiding the “Not Invented Here” or “reinventing the wheel”-attitude that is common to many Free Software and Open Source projects, and (ii) structuring the software that is produced in the project in modules, or components, of the appropriate size, and with a design that facilitates reuse (inside and outside of ROS and Orocos, which might be the two major frameworks being used in RoboHow.