Mobile Robots and Autonomous Vehicles

Course description

Mobile Robots are increasingly working in close interaction with human beings in environments as diverse as homes, hospitals, public spaces, public transportation systems and disaster areas. The situation is similar when it comes to Autonomous Vehicles, which are equipped with robot-like capabilities (sensing, decision and control).

Such robots must balance constraints such as safety, efficiency and autonomy, while addressing the novel problems of acceptability and human-robot interaction. Given the high stakes involved, developing these technologies is clearly a major challenge for both the industry and the human society.

This course is designed around a real-time decision architecture using Bayesian approaches. It covers topics such as:

• Sensor-based mapping and localization: presentation of the most popular methods to perform robot localization, mapping and to track mobile objects.
• Fusing noisy and multi-modal data to improve robustness: introduction of both traditional fusion methods as well as more recent approaches based on dynamic probabilistic grids.
• Integrating human knowledge to be used for scene interpretation and decision making: discussion on how to interpret the dynamic scene, predict its evolution, and evaluate the risk of potential collisions in order to take safe and efficient navigation decisions.

Course objectives

The objective of this course is to introduce the key concepts required to program mobile robots and autonomous vehicles. The course presents both formal and algorithmic tools, and for its last week's topics (behavior modeling and learning), it will also provide realistic examples and programming exercises in Python.

Who is this course for?

The course is primarily intended for students with an engineering or masters degree, but any person with basic familiarity with probabilities, linear algebra and Python can benefit from it.

The course can also complement the skills of engineers and researchers working in the field of mobile robots and autonomous vehicles.

Course outline

• Week 1: Objectives, hallenges, state of the art
• Week 2: Bayes and Kalman filters
• Week 3: Extended Kalman filters
• Week 4: Perception and situation awareness and decision making
• Week 5: Behavior modeling and learning (with examples and exercises in Python)

Pedagogical team

Authors:

• Christian Laugier, first class Research Director, Inria
• Agostino Martinelli, computer science researcher, Inria
• Dizan Vasquez, computer science researcher, Inria

Pedagogical support:

• Christelle Mariais, learning engineer, Inria Learning Lab.
• Isabelle Rey, learning engineer, Inria Learning Lab.

Additional resources

• Infographic : Figures from the 2nd session of the MOOC
• Videos and slides of the MOOC downloadable on Canal U

Partners

This MOOC was created by Inria’s MoocLab, as part of the uTOP project ( http://utop.fr/ – http://utop.inria.fr/). uTOP is an IDEFI project that aims to create a demonstrator for increasing research visibility through out training.