GMPE340 Kalman Filter and Sensor Fusion

Credits (ECTS):5

Course responsible:Jon Glenn Omholt Gjevestad

Campus / Online:Taught campus Ås

Teaching language:Engelsk, norsk

Course frequency:Every other year (odd-numbered years) from the academic year 2025/2026.

Nominal workload:125 hours.

Teaching and exam period:This course has teaching/evaluation in autumn parallel.

About this course

Autonomous vehicles provides the technology towards a more sustainable transportation. The performance of modern transportation systems has been greatly improved by the rapid development of connected and autonomous vehicles, of which high precision and reliable positioning is a key technology. This is one application that relies on the use of sensor fusion with the Kalman filter as the main workhorse providing reliable estimates of the system states (e.g. position, velocity and orientation).

This course introduces the use of stochastic processes and applied Kalman filtering with focus on positioning, navigation and timing applications (PNT).

The first part includes the essential notions of probability, an introduction to random signals and response to linear systems, state-space modelling and Monte Carlo simulations.

The second part contains the main theme of the course, which is applied Kalman filtering. This part starts with the basic filter derivation using the minimum mean square error approach. This is followed by various approaches to the base theory such as: the information filter, suboptimal analysis, conditional density viewpoint, Bayesian estimation, relationship to least squares adjustment (LSQ) and other estimators, smoothing and methods to deal with non-linearities.

Learning outcome

Students are to understand the basic filter derivation applied to dynamic systems. This is followed by various approaches to the basic theory such as: bayesian estimation, information filter, particle filter, least squares (LSQ) and other estimators. Further students are to understand methods for dealing with non-linearities.
  • Learning activities
    Lectures, colloquium, student presentations, exercises.
  • Teaching support
    Teaching support will be given primarily in connection with that part of the structured teaching that is set aside for exercise guidance. It will also be possible to communicate directly with the subject teacher by appointment during office hours.
  • Prerequisites
    Calculus and linear algebra. Differential equations. (MATH121/MATH111 and MATH122/MATH112) Parameter estimation. (GMPE240)
  • Recommended prerequisites
    Good programming skills (Python) (INF201 and INF202/INF203)
  • Assessment method
    3 hour final written examination. Grading A-F.

    Written exam Karakterregel: Letter grades Hjelpemiddel: C1 All types of calculators, other aids as specified
  • Examiner scheme
    The external and internal examiner jointly prepare the exam questions and the correction manual. The external examiner reviews the internal examiner's examination results by correcting a random sample of candidates exams as a calibration according to the Faculty's guidelines for examination markings.
  • Mandatory activity
    Exercises. Mandatory activities must be completed again when repeating the course.
  • Notes
    The course is recommended for students in Geomatics, Robotics and Data Science.
  • Teaching hours
    Lectures and discussion groups: 26 hours. Exercises: 52 hours.
  • Admission requirements
    Special requirements in Science