PhD thesis of Olivier Lebeltel (1999)
This thesis proposes an original method for robotic programming based on bayesian inference and learning.
This method formally deals with problems of uncertainty and incomplete information that are inherent to the field. Indeed, the principal difficulties of robot programming comes from the unavoidable incompleteness of the models used.
We present the formalism for describing a robotic task as well as the resolution methods. This formalism is inspired by the theory of the probability calculus, suggested by the physicist E T Jaynes: “Probability as Logic”. Learning and maximum entropy principle translates incompleteness into uncertainty. Bayesian inference offers a formal framework for reasoning with this uncertainty.
The main contribution of this thesis is the definition of a generic system of robotic programming and its experimental application. We illustrate it by programming a surveillance task with a mobile robot: the Khepera. In order to do this, we use generic programming resources called “descriptions”. We show how to define and use these resources in an incremental way (reactive behaviors, sensor fusion, situation recognition and sequences of behaviors) within a systematic and unified framework. We discuss the various advantages of our approach: statement of preliminary knowledge, taking into account uncertainty, direct and inverse programming. We suggest perspectives for our work: choice of architecture and planning. We place our work within a wider epistemological horizon while opposing, within the framework of autonomous robotics, the “traditional” approach concerning “high level cognition” and the “reactive” approach associated with the “low level cognition”. We finally show how our work proposes to establish a link between these two extremes.