Mobile Learning Robots - Topics

Learning approaches		Neural and fuzzy techniques
	If only a robot could learn to solve a task without having to program it explicitly. This could save a lot of development time. Moreover, this would allow non-expert users to use robots in daily life. The ultimate goal of our research is that a user should only specify or demonstrate what the robot should do, and not anymore how the robot should solve the task. [more]		Neural networks are structures that offer a flexible way to model arbitrary nonlinear functions by learning from data. Fuzzy controllers allow knowledge to be explicitly modelled in a human-like way. We can combine both paradigms in so-called hybrid networks that incorporate the best of both worlds. As an example, we have successfully developed non-linear (neuro-) fuzzy controllers for an assembly task. The ultimate goal of our research is that these (neuro-) fuzzy controllers can be synthesized automatically. [more]
Mobile manipulation		Shared autonomy
	One speaks of mobile manipulation when a mobile platform is equipped with one or more manipulator arms. This combination is a redundant system, meaning that movements and actions can be executed in more than one manner. A mobile manipulator should solve the problems and use possibilities provided by this redundancy in a smart way. [more]		The concept of shared autonomy appears in (mobile) robotics when both an intelligent system and a human operator are in control of the robot. An example is an intelligent wheelchair. The user performs the high-level task of planning and the intelligent wheelchair takes care of low-level fine motion control. [more]
Human-robot interaction		Programming by demonstration
	Human robot interaction is of great importance for the development of robots that operate outside production lines and cooperate with humans. It is necessary to develop techniques that allow untrained users to make efficient and safe use of their robot, using an intuitive and natural interface. This topic also involves research on emotional interaction with robots. [more]		Robots that cooperate with humans require easy programming methods allowing the inexperienced user to easily instruct the robot. Indeed, if robots are to become consumer products, their users cannot be expected to follow a formal programming procedure. [more]
Map building and navigation		Uncertainty and possibility theory
	This technique allows a mobile robot to automatically build a map of the environment based on the measurements it performs with its sensors. This newly built map then enables the robot to accuratelly localise itself in the environment.		Most problems in robotics arise because of some kind of uncertainty. Hence an important area of our research is dealing with this uncertainty. Besides the use of more classicial appraches such as probability theory and Bayesian reasoning, a recent line of research is possibility theory.
Cooperating robots, swarm behaviour and emergence
	Robot cooperation is an interesting and challenging research area. The capabilities of the combined system can extend those of the individual robots. In this topic, several applications have been developed.