View All Pages

Mobile Robotics

Format

The format of this third part of the course will be slightly different. We will make use of a so called "flipped classroom" setup. In addition to reading material like for the previous lectures you will also be provided with last year's recorded lectures and slides in advance of each session. You will be asked to complete a quiz before the session as before but what happens at the session will be different.

You have the material needed to pass the course already with the reading material and the recorded lectures and slides. The session together is meant to add something to that. As I want the sessions to be interactive and everyone feeling that they can be participate I will not record any part of the sessions. 

A rough outline of the session is

  1. I will start each session with a Q&A part, essentially asking if there are any questions on the lecture that I assume you already watched. So instead of having unanswered questions after the lecture I want you to bring those questions to our session together. This part will take no time if there are no questions.
  2. In the second part I will provide you with some small assignment(s),  connected to the material, to work on in groups. 

As before you will be able to attend over Zoom or from the lecture room. If you attend over Zoom you will work in breakout rooms. I expect you to have a camera so that the people that you interact with can see you. If you attend in real life be prepared to be within speaking distance from other people so that you can talk to each other.

My aim is to make the assignment(s) last max 45min . How long the session actually takes depends on the questions in the first part and the interaction in the second. Hopefully, the reason for taking longer will be that there is value created for you as a student.

 

Content

This part of the course covers mobile robotics. We will cover the topics

  • Lecture 9: Locomotion, Sensing and Perception
  • Lecture 10: Localisation
  • Lecture 11: Mapping and SLAM 
  • Lecture 12: Navigation

Most topics are covered in the Springer Handbook of Robotics. The book Probabilistic Robotics covers most of localisation and mapping in a textbook way (i.e. a bit more accessible than the handbook) but it is not complete in scope and give much more details than needed for what it does cover.

This part of the course spans a much larger area of robotics than the previous lectures and the amount of material to read if you wanted to read it all is therefore quite staggering. The aim for this part of the course is to get a basic understanding for the concepts and methods rather than going into details. So when reading, read with that in mind. 

First of all there are several ways in which we can implement a locomotion system. In the first lecture we will give an overview of different means of locomotion. When we start moving, the requirements on sensing becomes even larger. Even if the world is static, which we often assume, what the robot is exposed to will change because the robot changes position. We need sensing to sense the world around us so that we know how to interact with it, avoid running into it, etc. In the first lecture we will also look at different sensing modalities.

When we start moving in the world we introduce uncertainty not only in what the world around us looks like but also where we are in the world. Localisation is one of the enabling technologies, not only in mobile robotics, but also in, for example, location based services. In the second lecture we will look at the problem of localisation and will also discuss estimation in general.

In order to localise the robot we need to have a map of the environment to which we can relate our position. This map can come in many forms. When we use the GPS system the map can be said to consist of the GPS satellites and their known positions. For realise a truly autonomous systems, the system need to be able to build the map on its own. This is trick as we just said that the map is used for the localisation as well and the location is needed to do the mapping. In the fourth lecture we will look at the problem of SLAM, or simultaneous localization and mapping which is the solution to this chicken or egg problem. We will also discuss variations of the SLAM problem such as visual odometry.

In the final problem discuss the navigation problem, which refers to being able to go from A to B. This requires all of the above and also the ability to plan the motion and to execute it while making sure that obstacles are avoided. We will also, if time permits look at some challenges with building a complete system.

 

Assignment

In the assignment corresponding to this module you will build a map of the environment. This connects what we will talk about in all four lectures on mobile robotics. You will be using data from a laser scanner collected by a robot moving around in an unknown environment. The laser scanner allows the robot to gather information about the environment (sensing). It is using its encoders (sensing) to estimate the position while it is moving (localization). The robot uses the position information and the laser data to build a map of the environment (mapping). In the C-part we also discuss how to process this map to support motion planning and execution of such plans (navigation).