Title: Navigation Cost Mapping in Outdoor Environments from Air-Ground Vehicles Cooperation

With the growing interest in development of autonomous vehicles for outdoor environments, it is necessary to investigate techniques that support the autonomous navigation process. The autonomous navigation in outdoor environments has been widely explored by academic community, being analyzed all the factors that allow a safe and efficient displacement. For safe and efficient navigation, they are often considered obstacles in environment. However, unknown and unstructured surfaces may represent a crucial feature for the robot’s security or viability of the task. This work addresses the problem of autonomous mapping of navigation cost in external environments based on cooperation between ground and air vehicles. In this work were considered outdoor and free of obstacles environments, where difficulties can be found, such as: i) different types of surfaces; ii) roughness levels disparities; iii) and highly sloping surfaces. In this way, the main objective is to create quality navigation cost maps, maximizing the average confidence in navigation cost estimates. The methodology proposed in this thesis is divided into three main steps: i) Roughness level classification; ii) Navigation cost representation; iii) and Autonomous navigation for the environment. To validate the proposed steps, experiments were carried out with real robots, in different environments, in order to analyze the quality of the classification process, the navigation cost representation and the complete process of autonomous mapping. At the end, the results achieved at each proposed step are discussed.

Prof. Douglas Guimarães Macharet – Advisor (DCC-UFMG)
Prof. Mario Fernando Montenegro Campos – Co-advisor (DCC-UFMG)
Prof. Luiz Chaimowicz (DCC-UFMG)
Prof. Armando Alves Neto (DEE-UFMG)
Prof. Denis Fernando Wolf (ICMC-USP)