The recorded lectures and the written material of a course on Self-Driving Cars at the University of Tübingen have been made available for free. It's a first class opportunity to learn the in and outs of how to develop the software that powers self-driving cars.

Self-Driving is a topic of hot debate. Initially it had been heralded as the white knight of the big city that would decongest the streets, lower the risk of accidents, provide mobility for elderly and people with disabilities and decrease pollution for a more healthy environment. But after crashes, fatalities and malfunctions, some said that a fully autonomous cars might just be a dream that will never come to fruition. After all building the brains for such version of a car is not the easiest thing in the world.

Fear not though; this course taught at the University of Tübingen by Prof. Andreas Geiger has been made freely available to everyone in an attempt to collectively push things forward.

And is taught under a novel approach too, that of the 'flipped' classroom. This means that the lectures are provided via YouTube and must be watched before the respective interactive live sessions. Interactive live sessions is where questions regarding the lecture and exercises are posed and discussed together.

As such the recorded sessions and the related material have been made available;the interactive ones are privilege of the College's students attending the class on premises.

The goal of the course it to develop an understanding of the capabilities and limitations of autonomous driving solutions and gain a basic understanding of the entire system comprising
perception, planning and vehicle control. But before jumping in, it's important to jot down the prerequisites for attending:

• Basic Computer Science skills: Variables, functions, loops, classes, algorithms
  
  
• Basic Python and PyTorch coding skills
  
  
• Basic Math skills: Linear algebra, probability and information theory
  
  
• Experience with Deep Learning (eg. , through participation in our Deep Learning lecture)

The topics discussed from a birds eye view are:

• History of self-driving cars
  
  
• End-to-end learning for self-driving (imitation/reinforcement learning)
  
  
• Modular approaches to self-driving
  
  
• Perception (camera, lidar, radar)
  
  
• Localization (with visual and road maps)
  
  
• Navigation and path planning
  
  
• Vehicle models and control algorithms

That list might give the impression of something too complicated, but at the most basic level self driven cars rely heavily on cameras to observe the road and its obstacles, with some models maybe even carrying at least 20+ cameras. Secondly, what's needed is the LiDAR which is sensors that fire pulses at objects and record the time it takes to bounce off those objects in order to create a 3D map of the surroundings of the car.

And finally it's how those cars learn, which poses as the starting point of the course after the Introduction, in Lecture 2-Imitation Learning: Approaches to Self-Driving. That is also the most important lecture in order to get a grasp of the context. The rest of lectures just build on the fundamentals:

Lecture 3 is about Direct Perception
Lecture 4 Reinforcement Learning
Lecture 5 Vehicle Dynamics
Lecture 6 Vehicle Control
Lecture 7 Odometry, SLAM and Localization
Lecture 8 Road and Lane Detection
Lecture 9 Reconstruction and Motion
Lecture 10 Object Detection
Lecture 11 Object Tracking
Lecture 12 Decision Making and Planning

The crisp material is further enhanced by the Professor's high level of conveyance, which wraps up a fine course on a state of the art topic. Plus it's been taught by a respectable University and also free! What else could you ask for?

More Information

Self-driving cars Youtube playlist

Main University site

Related Articles

Artificial Intelligence, Machine Learning and Society

To be informed about new articles on I Programmer, sign up for our weekly newsletter, subscribe to the RSS feed and follow us on Twitter, Facebook or Linkedin.

Comments

Make a Comment or View Existing Comments Using Disqus

or email your comment to: comments@i-programmer.info