Autonomous Systems

Objectives

Fundamental concepts involved in systems composed of diverse physical agents are covered, with diverse autonomy degrees (sensors, processors, actuators, robots) spatially distributed. Fundamental concepts and methods for selflocalization under uncertainty on the observation and motion models are described. Methods for integrating the information from multiple sensors are presented, for positioning and for representing the world map where the sensors are situated, as well as methods for problem solving in cooperative systems, including cooperative perception, and task assignment, planning, and coordination. Fundamental concepts on functional, software, and hardware architectures concludes the course.

Program

1. [2h] Introduction to autonomous systems: mobile robots, mobile and static sensor networks. Uncertainty in robotics.
2. [4h] Probabilistic representation of uncertainty: probabilistic models of observation and action. Bayesian inference.
Bayes filter and its particular cases.
3. [6h] Bayesian localization.
4. [1h] Probabilistic occupancy grid mapping.
5. [3h] Simultaneous localization and mapping (SLAM).
6. [6h] Task planning: classical planning; planning under uncertainty: Markov decision processes (MDP). Reinforcement learning.
7. [2h] Plan representation and its execution coordination. Performance analysis.
8. [3h] Cooperative systems: cooperative localization and tracking of objects. Sensor integration: distributed sensor fusion methods. Cooperative task assignment, planning, and coordination.
9. [1h] Functional, software, and hardware architectures.

Teaching Methodologies

Group project involving course topics, with weekly progress presentations, final report, and poster presentation (70%) + Individual written exam covering the course program (30%).

Bibliography

Probabilistic Robotics, S. Thrun, W. Burgard e D. Fox, 2005, MIT Press; Planning Algorithms, Steven Lavalle, 2006,
Cambridge University Press; Reinforcement Learning: an introduction, R. Sutton and A. Barto, 1998, MIT Press;
Probabilistic Robotics, Sebastian Thrun, Wolfram Burgard and Dieter Fox, 2005, MIT Press. http://www.probabilisticrobotics.org/; Artificial Intelligence: A Modern Approach (chaps. 7, 8, and 10), Stuart Russell and Peter Norvig, 2009,
Pearson.; Reinforcement Learning: An Introduction (chaps. 3 and 6, Richard S. Sutton and Andrew G. Barto, 2018, MIT
Press; Springer Handbook of Robotics (chaps. 35 and 53), Bruno Siciliano and Khatib Oussama, 2016, Springer.