Hoppa till huvudinnehållet

EL2320 Applied Estimation 7,5 hp

Course memo Autumn 2021-50179

Version 3 – 10/20/2021, 11:32:36 AM

Course offering

Autumn 2021-1 (Start date 01/11/2021, English)

Language Of Instruction

English

Offered By

EECS/Intelligent Systems

Course memo Autumn 2021

Course presentation

This course mainly focuses on estimation using Kalman filter and particle filter. In this course you will go into depth and get very familiar with two methods that in the other course you will learn are specific examples of  wider classes of methods.

Course Structure

There are 12 lectures. Both theory and practice of estimation will be covered. Getting practical skills in anything requires you to get hands-on experience and as such the work between the lectures will be very important.

Two labs cover the Kalman and particle filters. 

For the final project, the student should work in pairs and implement an estimation method. Each student needs to write an individual report including a literature study. 

Recommended Prerequisits:

Courses corresponding to SF1624 Algebra and Geometry, SF1901 Probability Theory and Statistics,

Literature

The official course book is "Probabilistic robotics" by Thrun, Burgard and Fox, The MIT Press, ISBN 0-262-20162-3. Some of the material will be in the form of research publications. The students are assumed able to research for additional material to solve the project assignment.

Headings denoted with an asterisk ( * ) is retrieved from the course syllabus version Spring 2019

Content and learning outcomes

Course contents

The course focuses on giving the participants practical experience in using different estimation techniques on real problems. Examples used in the course are, for example, from navigation with mobile robots.

The following will be covered in the course: Observability, the Markov assumption, data association, estimation techniques such as Kalman filter, extended Kalman filter, particle filter, Rao-Blackwellized particle filter, Unscented Kalman Filter

Intended learning outcomes

The overall goal of the course is to give the participants theoretical as well as practical skills and experience in estimation. The course will start from a number of concrete examples to motivate the need for various filtering techniques such as Kalman filters and particle filters.  After completing the course the student should:

  • be able to describe the parts of a Bayesian recursive filter in terms of the underlying probabilities, compare and contrast different estimation techniques, and select and apply appropriate techniques to problems. 
  • have reflected on the relationship between measurement uncertainty, probability theory and estimation methods.
  • have gained experience in finding information from current scientific literature including recently published journal articles. As well as presentation of results in well structured scientific reports.

Learning activities

There are 12 Lectures, two Labs, a project, and an exam.  The labs are on the Extended Kalman Filter and the Particle Filter.  They go thruough an example in fine detail.  Lab reports are uploaded and graded pass or fail individually.   

The project  serves two goals, one is to give a deeper understanding of estimation and the other is to give students experience writting a scientific report.  The students can chose to do a literature study on some topic which gives a maximum project grade of C or they can add a simple implementation of there own chossing for a grade up to A.

The exam is graded on the scale P-F.

The project grade to gives a final grade A-F. 

Detailed plan

Learning activities Content Preparations
   
   
   


Schema HT-2020-AE20

Covid restrictions

As of Sept. 29 KTH and the government have lifted the restrictions on social distnacing.  The course will therefore be given with in person lectures in the scheduled rooms but  in-person examination in January is not allowed.   We will therefore have a timed home assignment during the time allocated for the exam and a follow-up oral examination of the answers uploaded. 

We may add help sessions over Zoom or in person tba.  There are in particular two occassions in which physical presense is required to gain bonus points on the Exam.    

 

Preparations before course start

Recommended prerequisites

Courses corresponding to SF1624 Algebra and Geometry, SF1901 Probability Theory and Statistics, SF1635 Signals and Systems, part I. Being able to program in MATLAB.

Literature

Probabilistic Robotics by Sebastian Thrun, Wolfram Burgard, and Dieter Fox.

Examination and completion

Grading scale

A, B, C, D, E, FX, F

Examination

  • PRO1 - Project, 2.0 credits, Grading scale: P, F
  • PRO2 - Project, 2.0 credits, Grading scale: P, F
  • TEN1 - Examination, 3.5 credits, Grading scale: P, F

Based on recommendation from KTH’s coordinator for disabilities, the examiner will decide how to adapt an examination for students with documented disability.

The examiner may apply another examination format when re-examining individual students.

The basic part of the examination in the course consists of two lab assignments (PRO1), a project  (PRO2) and an exam (TEN1). These are credited as,

PRO1: 2.0hp 
PRO2: 2.0hp 
TEN1: 3.5hp   

Each of these will be reported to the system with a P/F grade and passing them means that the student has passed the course. The final grade is then an assessment based on the level above passing on these moments as described below.

The final grade for the course will be an average of the grade on the exam and the implementation project. This project assignment can be completed in groups of two students but each needs to write a separate report and will be assessed separately.  

Not opting to do an implementation project (ie.  only doing  a literature study)  thus gives a final grade of half the exam grade.

The section below is not retrieved from the course syllabus:

Project ( PRO1 )  This is based on two labs.

Project ( PRO2 )  This is based on the individual written project report

Examination ( TEN1 ) Pass Fail Exam

Other requirements for final grade

To get a passing grade in the course the students need to pass the labs, the mandatory part of the project assignments and the exam.

Grading criteria/assessment criteria

Passing labs are required to pass the course.   Labs can give bonus points.

Passing the exam is required  to pass the course.

The exam is graded out of 50 and 25 is passing but one must also pass each of three subparts of the exam.  Passing a subpart requires more than 50% of the points for that part.

The project implementation is not required but if you do not do  a coding project you still are required to upload a report by the project deadline.  The report can be either a literature review where you  summarize  a published scientific paper on a relevant topic along with comparisons to other papers or describe an implementation of an estimation  method of your choosing.   

The project or literature reports are given a letter grade.  A project report can get grades A-F while a literature report gets C-F.

This table has the project grade as the first row and the bonus points as the first column.  The final grade can then be found in the body.

Final Grade Table:

Bonus\Project Grade A B C D E
                               6 A A B D E
                               5 A B B D E
                               4 A B B D E
                               3 A B C D E
                               2 A B C D E
                               1 A B C D E
                               0 A B C D E
 

For the project  grade there are three types of criteria, Amount, Clarity, and Insight.    For a literature study it is only clarity. 

So you get one point for doing adequate 2 for a little extra and sometimes 3 for a lot more.

Implementation (The programming part, 5 pts.):

Amount 1-3 points depending on how extensive the coding part was (per person so three can do more than one).   Here there is a maximum amount that you can get credit for so do not over do the implementation.   

Analysis 1-2 points depending on how much you do after the implementation of the method.  So more simulations, more types of figures generated, more parts of the results analysed.  Again once you have gone above average you gain nothing by simple adding figures.

Report (7 pts.):

Organization  2: pts Does it read well.  Are you motivating what you did or just stating what you did.  Can I really understand what you did and that you understood.

Background and References 3 pts.  

Did you give the context well?  Are the references used well in the texts to motivate and add credibility to your report.  This means you show that you understand how the state of the art has gotten to where it is and what were the significant steps.  So if you simple throw the citations in you get 1 point, using them to support the text and arguments, motivation can get up to 3.  

Method 2 pt.  Is the method well explained.    If I can figure out what you did with effort 1 pt. If you manage to explain both at a high level (what is the strategy of the method) and a low level (enough to reproduce, NOT code that is too low).  Remember readers that want to can code they need to understand what is being done. 

Insight  (2 pts):

Analysis 1 p For something special in the experiments and their analysis.  That very beautiful graph or movie that shows exactly how this works...

Background 1 p For something special in describing the state of the art and theory.

The grade then is based on a 2D table lookup for the total of Amount and Clarity with the Insight points if any moving your grade one or possibly 2 grades up.  So Report grade across the top row and Implementation down the first column. 

 

  1 2 3 4 5 6 7
0 F F F E E D D
1 F F F E E D C
2 E E D D C C C
3 E D C C C C C
4 E C C C C C B
5 D C C C C B B

Note that a literature study has no implementation so you are on row 0 of the table and can only get a C byt earning 6 or 7 points for 'Report' and one  'Insight for background' point.  The method would be more about explaining the method(s) of a reference paper.

Opportunity to complete the requirements via supplementary examination

Fx

To obtain an Fx for the final course grade one must have passed the exam and done all the other assignments  in the course and passed all but one.  The one that was not passed needs to have been close to passing.  Then one can complete that after the course is finished.    

Ethical approach

  • All members of a group are responsible for the group's work.
  • In any assessment, every student shall honestly disclose any help received and sources used.
  • In an oral assessment, every student shall be able to present and answer questions about the entire assignment and solution.

The section below is not retrieved from the course syllabus:

Information on how to approach collabortion is given in the first lecture.  Slides are in Canvas.

Further information

No information inserted

Course Coordinator

Teachers

Examiner

Round Facts

Start date

1 Nov 2021

Course offering

  • Autumn 2021-50179

Language Of Instruction

English

Offered By

EECS/Intelligent Systems

Contacts

Course Coordinator

Teachers

Examiner