The challenge to develop autonomous mobile manipulators that are able to transport large size objects in cooperation with humans is very motivating since such robots would be very useful in many fields related to our daily activities, such as in construction sites, at home, at office or at industrial plants. Thus, many researchers concentrate their efforts in the development of human-robot cooperation systems and have proposed control algorithms for a robot or teams of robots handling a single object in cooperation with humans (e.g. [4], [5], [6] and [7]). Despite the efforts, no final and robust solutions exist. From the point of view of the cooperating robot the environment, which consists of the manipulated object, the human and the world scenario (static or dynamic), exhibits complex dynamic behavior. The problem is exacerbated when the environment is unknown and no path is given to the robot.

Here, we address the problem of controlling and coordinating the movement of an autonomous mobile robot that must help a human to carry large size objects in an unstructured indoor environment. Particular to work, we use non-linear dynamical systems as a design and theoretical tool to design a control architecture that controls the behavior of the cooperating robot. Specifically, the time course of the control variables is obtained from attractor solutions of dynamical systems.

We will assume that the robot has no prior knowledge of the environment and we choose a human-follower motion control strategy. The human grasps one end of the object and walks from an initial position to a final target destination. The robot (i.e. follower ) grasps the other end of the object, takes the human as a reference point, and must steer so as to keep at all times the correct orientation and distance to the human, that permits it to cooperate with the human in the transportation task, and simultaneously avoid any obstacles that may appear.

The control architecture for the robot will be structured in terms of elementary behaviors. The individual behaviors and their integration will be generated by nonlinear dynamical system [2]. The benefit is that the mathematical properties associated with the concepts enable system integration including stability of the overall behavior of the autonomous system.