PhD Alumni

Basilico Nicola

Present position:

Thesis title:  Navigation Strtategies for Exploration and Patrolling with Autonomous Mobile Robots
Advisor:  Francesco Amigoni
Research area:  Artificial intelligence, robotics and computer vision
Thesis abstract:  
Recent advances in mobile robotics showed that the employment of autonomous mobile robots can be an effective technique to deal with tasks that are difficult or dangerous for humans. Examples include exploration, coverage, search and rescue, and surveillance. Fundamental issues involved in the development of autonomous robots span locomotion, sensing, localization, and navigation. One of the most challenging problems is the definition of navigation strategies. A navigation strategy can be generally defined as the set of techniques that allow a robot to autonomously decide where to move in the environment in order to accomplish a given task. As a typical example, consider a robot exploring and mapping an unknown environment that has to select the next location, within the currently explored portion of space, where to take a sensing action. Independently of the particular applicative scenario, navigation strategies have a remarkable influence over the performance of the task execution and significantly contribute in building the robot's autonomy. Despite their centrality, a general characterization of navigation strategies and the definition of application-independent methods for their development are still largely considered as open issues. The majority of works proposed in literature provide ad hoc approaches, making the proposed techniques hardly adaptable to scenarios different from that they have been tailored for.

In this dissertation, we aim at contributing towards a general framework for navigation strategies. Our approach is based on considering a mobile robot as a decision maker that makes decisions about where to move. This allows us to study the definition and the adoption of general decision-theoretic techniques for defining navigation strategies. We apply this approach to relevant applicative domains that are classified according to some dimensions, e.g., single or multi robot, partial or global knowledge of the environment. The first case we address involves exploration for map building of unknown environments and search and rescue for victims. To deal with these settings a technique called Multi Criteria Decision Making (MCDM) has been applied. In MCDM a robot evaluates the candidate locations in a partially explored environment according to an utility function that combines different criteria (for example, the distance of the candidate location from the robot and the expected amount of new information acquirable from there). Criteria are combined in a general utility function that accounts for their synergy and redundancy. In the second case we consider robotic patrolling, where a mobile robot navigates through an environment to detect possible intrusions. The approach we propose to compute effective patrolling strategies is based on modeling the patrolling setting as a competitive game between the patroller and the intruder. The optimal patrolling strategy is thus determined by computing an equilibrium of the game.

The obtained results are encouraging and suggest the possibility of developing a general theoretical framework in which navigation strategies can be defined.