The field of robotics is, by essence, multidisciplinary, because it involves skills of numerous disciplines such as mechanics, mechatronics, electronics, automation, computer science and artificial intelligence. The interdisciplinary, rich in its exchange and interactions, stands out naturally as a methodological approach on robotic issues and has driven my scientific approach from my early days as a robotics researcher between cognitive sciences and artificial intelligence, automatics and computer science.
Within this context I defended my doctoral thesis in Computing Science the 11th of january 2017 entitled :
" The RHIZOME Architecture: A Hybrid Neurobehavioral Control Architecture for Autonomous Vision-based Indoor Robot Navigation "
I received my Ph.D. degree from the judging panel composed by the following researchers:
Doctoral Directors:
Arnaud Revel Professor, L3I University of La Rochelle
Michel Ménard Professor L3I University of La Rochelle
Jury:
Patrick Hénaff Professor, LORIA, École de Mines de Nancy
Olivier Strauss Associate Professor HDR, LIRMM, University of Montpellier
Emmanuelle Grangier Ph.D, UFR Intermédia, University of Toulon
Jean-Philippe Domenger Professor, LABRI, University of Bordeaux
Philippe Gaussier Professor, ETIS, University of Cergy Pontoise
Armelle Prigent Associate Professor, L3I, University of La Rochelle
You can download the dissertation here
Abstract of the thesis:
The subject of my thesis is a contribution to the problem of autonomous indoor vision-based mobile robot navigation, which is still a vast ongoing research topic. It addressed it by trying to conciliate all differences found among the state-of-the-art control architecture paradigms and navigation strategies. Hence, I proposed the RHIZOME architecture (Robotic Hybrid Indoor-Zone Operational ModulE): a robotic control architecture capable of creating a synergy of different approaches by merging them into a neural system. The interactions of the robot with its environment and the multiple neural connections allow the whole system to adapt to navigation conditions.
The RHIZOME architecture emerged out of the will to provide an adequate autonomy to mobile robots allowing them to navigate within an environment while being capable of adapting themselves to unforeseen situations presented in it. It consists of a behavior-based hybrid architecture that fuses the a priori information and real-time visual information of the world into a neural structure.
The a priori information of the world is used to only corroborate the real-time visual information perceived during navigation, contrary to most hybrid architectures that use it to directly control the actions of the robot. Additionally, instead of using a complete motion path, the RHIZOME architecture makes use of artificial navigation signs and their expected sequence in the navigation path. Such sequence is provided to the robot either by means of a program command or by enabling it to extract itself the sequence from a floor plan. This latter implies the execution of a floor plan analysis process. Consequently, in order to take the right decision during navigation, the robot is able to process both set of information, compare them in real time and react accordingly. When the navigation signs are not present in the navigation environment as expected, the RHIZOME architecture allows the robot to learn and recognize places based on natural navigation signs that it perceives in the environment. Thus, the robot is still able to achieve its final destination by overcoming the unforeseen situations.
The RHIZOME architecture is composed of a hybrid behavioral structure that combines a deliberative module and one or several behavioral modules.Thanks to the generic composition of the proposed RHIZOME architecture, it is possible to develop the architecture further with respect to robustness and completeness by building new layers and modules separately and simply adding them without modifying the already in-built components or modules. One can certainly go further and further on the construction of many modules, as long as there are always new scenarios constraints to overcome.
In the context of this work, the RHIZOME architecture was conceived, built and implemented through three different scenarios under which, three interdependent architectures emerged, each responding to the different scenario constraints.
· Rhizome1: Deterministic scenario – Exploring the world with little information
· Rhizome2: Deterministic scenario – Map-dependent autonomous navigation
· Rhizome3: Stochastic scenario – Self-learning and adapting
The overall architecture has been implemented on the NAO humanoid robot. Real-time experimental results during indoor navigation under both, deterministic and stochastic scenarios show the feasibility and robustness of the proposed unified approach.
The following video shows the robot navigation under a deterministic scenario where the robot uses a floor plan of the building to navigate to its final destination.
Keywords: Artificial neuronal network-based control architecture, autonomous mobile robot indoor navigation, visual perception, data merging, floor plan analysis, pattern recognition, hybrid behavior-based approach.
Additional Research Activities
Plenary talk at the national workshop:
During my doctoral research project, I had the opportunity to assist, participate and present my research work in different events. From demonstrations of the robot while navigating in unknown environments to nonspecialized public, to presentations to researchers of the same field (neural networks, robotics, and artificial intelligence):
JOURNAL REVIEWS