EG504N: LOCALISATION AND MAPPING IN THE INDUSTRIAL DOMAIN (2021-2022)

Last modified: 31 May 2022 13:05

Course Overview

The aim of the course is to give an overview of the key techniques for enabling mobile robots to localise themselves, map their environments or do both simultaneously. The course introduces students to the fundamentals of statistical modelling and state estimation, widely used in automated vehicles and industrial automation.

Course Details

Study Type	Postgraduate	Level	5
Term	First Term	Credit Points	15 credits (7.5 ECTS credits)
Campus	Aberdeen	Sustained Study	No
Co-ordinators	Dr Pascal Meissner

What courses & programmes must have been taken before this course?

Master Of Science In Industrial Robotics
Any Postgraduate Programme

What other courses must be taken with this course?

None.

What courses cannot be taken with this course?

None.

Are there a limited number of places available?

Course Description

The course is an introduction to the paradigm of probabilistic robotics, applied in the context of mobile robotics. It covers the SLAM problem and its various forms as well as the different variants of recursive Bayesian filtering (e.g., the KF and EKF) and their underlying assumptions. The course begins with a recap of the basic concepts of random variables, probability distributions, conditional independence and Markov assumption. Then, the characteristics of probabilistic motion and sensor models for mobile robots are introduced just as the differences and similarities between Occupancy Grid Maps and Counting. An overview on the characteristics and designs of landmark- and grid-based SLAM solutions is given, including the effects of Rao-Blackwellization. Additionally, the basic python commands to implement Bayes filters and SLAM systems are conveyed.

Course Content

Introduction – Nomenclature, history, state of the art, challenges, course logistics
Probability Review – Events, axioms of probability, independent events, Bayes Rule, Bayes Filter
Probabilistic Modelling – Odometry- and velocity-based motion models, beam- and scan-based sensor models
Localisation with Nonparametric Filters – Discrete Bayes Filter, importance sampling, particle filter
Localisation with Gaussian Filters – Kalman Filter, extended Kalman Filter
Mapping with Known Poses – Occupancy maps, reflection probability maps
Landmark-based SLAM – SLAM problem, EKF SLAM, loop closing, Rao-Blackwellization, FastSLAM
Grid-based SLAM – Scan matching, FastSLAM, improved proposals, selective resampling

Contact Teaching Time

Information on contact teaching time is available from the course guide.

Teaching Breakdown

More Information about Week Numbers

Details, including assessments, may be subject to change until 30 August 2024 for 1st term courses and 20 December 2024 for 2nd term courses.

Summative Assessments

First Attempt

1x Lab Report: Individual (15%)
1x Take Home Exam (15%)
1x Online Open-Book Exam (70%)

Alternative Resit Arrangements

1x Online Open-Book Test (100%)

Formative Assessment

There are no assessments for this course.

Course Learning Outcomes

Knowledge Level	Thinking Skill	Outcome
Procedural	Apply	Apply Bayes (filter) formulae and sample from probability density functions
Procedural	Apply	Determine and apply probabilistic sensor and motion models
Procedural	Apply	Implement realizations of Bayes filters and compute location estimates for robots
Procedural	Analyse	Build and analyse grid maps
Procedural	Apply	Determine solutions to data association problems
Factual	Analyse	Prove properties of basic concepts from probability theory
Procedural	Analyse	Differentiate between localisation and SLAM systems as well as outline auxiliary techniques for SLAM solutions
Conceptual	Analyse	Discuss the steps and components of realisations of Bayes filters
Procedural	Evaluate	Assess and implement components of landmark- and grid-based solutions to the SLAM problem