Robotique

Autonomous robotics

Description: This course will introduce the field of autonomous robotics (vehicle driving, exploration and inspection robots, etc.) by showing how this issue integrates a wide range of technologies (localization (SLAM), point clouds, planning, pattern recognition) and how this integration is achieved at the system level (illustrations with ROS). The laboratory work associated with the course will be carried out on the robots of the Metz campus robotics platform and in simulation. This work will provide an opportunity to integrate different machine learning and signal processing techniques on robots moving in their environment. The course and practical applications will allow students to discover these techniques in real-world cases.

Content: 8 practical sessions covering the entire process from discovering simulation environments and programming with ROS to implementation on real robots. Topics covered include localisation, mapping, planning and trajectory execution.

Prerequisites: It is necessary to have a good knowledge of Python programming and a sound knowledge of probability. The practical work also requires a minimum level of familiarity with Linux.

Learning outcomes: At the end of this course, students will be familiar with ROS and the key concepts of autonomous robotics. They will have experienced the difficulty of coupling information processing with a system operating in a real environment and will have gained experience in implementing solutions, guided by the methodology induced by ROS.

Teaching methods: The lectures introduce the essential concepts and algorithms for autonomous robotics (13.5 HPE), supplemented by tutorials (3 HPE) and extensively supported by practical work in simulation and on real platforms (22.5 HPE).

Means: Tutorials, consisting of exercises, will enable students to apply the concepts covered in class. Practical work will enable students to programme and test the algorithms covered in class in simulations and on real robotic platforms using the ROS2 framework.

Evaluation methods: 2h written exam, can be retaken

Course supervisor: Francis Colas

Geode ID: SPM-INF-017

CM:

  1. Intro (1.5 h)
  2. Introduction à ROS (1.5 h)
  3. Rappels de probabilités (1.5 h)
  4. Estimation d’état (1.5 h)
  5. Localisation (1.5 h)
  6. Carto + SLAM (1.5 h)
  7. Planif (1.5 h)
  8. Navigation (1.5 h)
  9. Architecture et interaction (1.5 h)
  10. Robotique et réseaux de neurones (1.5 h)

TD:

  1. Filtres de Kalman (1.5 h)
  2. Localisation (1.5 h)

TP:

  1. ROS et simulation (3.0 h)
  2. Filtre de Kalman et estimation d’état (1.5 h)
  3. Localisation (3.0 h)
  4. Carto + SLAM (3.0 h)
  5. Path planning (3.0 h)
  6. Path following (3.0 h)
  7. Navigation et robots réels (3.0 h)
  8. Integration (3.0 h)