• 090943 - COGNITIVE ROBOTICS

Aims and learning outcomes

This course addresses the methodological aspects of Cognitive Robotics. Cognitive Robotics is about endowing the robot with intelligent behaviour by designing and deploying a processing architecture making the robot apt to  adapt and learn, allowing it  to decide how to attain complex goals in a complex world. Perception and action, and how to model them in neural and symbolic representations are therefore the core issues to address.

Inspiring models of Cognitive Robotics arise from different disciplines: the neural architectures from neuroscience, the basic behaviours from etology, motivations and emotions from psychology, the multirobot behaviour from sociology. Those models could be implemented in terms of formal logic, probabilistic, and neural models.

Implementation issues are approached and developed into an integrated middleware for robotics, to give students the experience of a quite professional way to develop and experiment robotics algorithms.

The last part of the course will focus on applying the knowledge from the initial lectures to the key approaches to robot design and control. The diversity of robot research provides an enormous number of technical papers, from which a few case studies are selected and discussed.

Syllabus

1.  Deliberative systems for cognitive robots

• reactive, cognitive, hybrid architectures.
• discrete and continuous planners: action representation and map representation.
• non linear planners, partial order planners, Graphplan, SATplan, PDDL (Planning Domain Definition Language).
• random planning in C-space (Probabilist Road Map, SBL, RRT).
• localization and mapping: probabilistic models, sensor models, SLAM.

2.  Bioinspired controllers for autonomous robots

• neural controllers and neural models of space and paths.
• neural models for the hand-eye coordination.
• evolutive systems for learning and control.
• imitation learning.
• collective control of actions and perceptions.

3.  User/robot and robot/robot interaction

• the environment as a communication medium: distributed sensing for robot/robot interaction.
• multisensorial interfaces: physical principles and telecontrol.
• bodily interface to interact with real and virtual worlds.
• case studies in prostheses and rehabilitation robotics using EMG signals.

4.  Constructing effective physical agents

• morphologic robotics.
• bioinspired systems.
• new actuators and sensors for cognitive robots.
• case studies of robots with advanced mobility (legged, swimmers, etc).

5.  Middleware and system integration

• the ROS (Robot Operating System) environment for robot simulation and control.
• robot models in ROS.
• sensor integration in ROS.

Prerequisites

Prerequisites include basic principles of robotics terminology and building blocks.

The final examination is a written test. Project work is also required, either in form of a collective workshop or as a personal technical homework.

provides the text of lectures and a reading list of downloadable teaching material