This work takes inspiration from how human infants gradually acquire knowledge about how their body works by performing suitable exploratory actions. With the aim of replicating this important feature also in robotic systems, in this paper we propose a method that combines adaptive robot control, meant to determine both the kinematic parameters and the type of joint (prismatic or revolute), and trajectory optimization, meant to determine the persistent excitation motions required to acquire the

more »

... maximum amount of knowledge about the robot itself. The ultimate aim is to enable robots to learn and become aware of their physical-mechanical structure. This would favour the application of powerful and predictable model-based control techniques even when only a minimal knowledge of the robot is available (i.e. the number of joints in our case). A preliminary validation is carried out through simulations on simple 1 DoF and 2 DoF robots, showing promising results and potential improvements to be addressed in the future.