Carla Koike

Bayesian Approach to Action Selection and Attention Focusing

Autonomous sensory-motor systems, situated in dynamic environments, must continuously answer the ultimate question: how to control motor commands knowing sensory inputs? Solving this question is a very complex problem, because a huge flow of information must be treated under several restrictions: real-time constraints, bounded memory space, and limited processing power. One additional and major challenge is to deal with incomplete and imprecise information, usually occurring in dynamic environments. In this thesis, we address the problem of controlling autonomous sensory-motor systems and propose a succession of cumulative hypotheses and simplifications. They are defined within a precise and strict mathematical framework, called Bayesian programming, an extension of Bayesian networks. This succession consists of five stages: Utilisation of internal states; First-order Markov assumption, stationarity and Bayesian filters; Exploiting partial independence; Addition of behaviour selection mechanism; and addition of Attention focusing. The validity of these hypotheses can be discussed and challenged. However, the corresponding inferences and resulting decisions are derived mathematically. Each description of a stage is followed by its analysis according to memory requirement, processing complexity, and difficulty of modelling. Further discussions regarding robot programming and cognitive modelling points of view are also presented. Finally, we describe an implementation on a mobile robot. The results demonstrate that the proposed framework is adequate for practical purposes.

A Bio-inspired Apodal and Modular Robot

Reconfigurable modular robotics is based on the use of autonomous units that can be attached in many ways to exert different kinds of functions. The main idea consists on the robots capacity to adapt to unexpected environmental circumstances and execute its mission. The current works goal is the adaptation of the mechanical system for the apodal movement, based on bio-inspired configuration of a snake for search and rescue situations on debris. Some aspects of the movement are described, followed by the definition and development of the robots structure, aiming for low cost production and easy replacement of the modules. Simulations and experimental tests were conducted to evaluate the robots performance.

Deformable Virtual Zone and Particle Filters Applied to Obstacle Avoidance in Mobile Robotics

Obstacle avoidance in a control system of an autonomous mobile robot is essential for assuring safety. Complete models, or maps, can be used to provide position of fixed obstacles, but in an unstructured environment, sensors are imperative to perceive mobile or unexpected obstacles and, consequently, be able to react to their presence. This article presents a novel technique for obstacle tracking inside a Deformable Virtual Zone (DVZ) using Particle Filtering to help keep the robot aware of an obstacle presence in its surroundings even after it is no more detected by its sensors.