Om kursen
This course deals with methods for solving various types of tracking problems. We can roughly divide the content into the following areas, with the corresponding techniques listed after:
Aim
To obtain a good theoretical and practical knowledge of a general tracking system.
Objectives
By the end of this course the participants will be able to use measurements to acquire knowledge regarding the states of a set of objects (targets). interpret the output from a tracking system and use it to make decisions. formulate sensor and motion models that can be utilized to perform tracking. implement and understand the commonly used linear and non-linear filtering algorithms. select a suitable data association technique among the frequently used ones. implement different methods to detect new objects and to initiate their tracks. design and implement a complete tracking system in simplified examples.
Organization
We will meet for one three hour session every week. The participants are expected to solve problems and implement algorithms every week. During the lectures we will discuss the last week's homework and material. At the end of each meeting I will present next week's topic and specify the homework.
Examination
To pass the course you are required to complete a minimum of 80 percent of the homework assignments.
Prerequisites
A decent background in topics such as statistics, estimation theory, and stochastic processes is useful.
- Linear filtering: the Kalman filter and particle filters.
- Non-linear filters: extended Kalman filters, unscented Kalman filters and particle filters.
- Multiple model filters: Interactive Multiple Models (IMM), multiple model pruning and change detection using CUSUM.
- Data association: Nearest Neighbors (NN), Probabilistic Data Association (PDA), Global Nearest Neighbors (GNN), Joint Probabilistic Data Association (JPDA) and Multiple Hypothesis Tracking (MHT).
- Track handling: Sequential Probability Ratio Tests (SPRT) and "N of M".
Aim
To obtain a good theoretical and practical knowledge of a general tracking system.
Objectives
By the end of this course the participants will be able to use measurements to acquire knowledge regarding the states of a set of objects (targets). interpret the output from a tracking system and use it to make decisions. formulate sensor and motion models that can be utilized to perform tracking. implement and understand the commonly used linear and non-linear filtering algorithms. select a suitable data association technique among the frequently used ones. implement different methods to detect new objects and to initiate their tracks. design and implement a complete tracking system in simplified examples.
Organization
We will meet for one three hour session every week. The participants are expected to solve problems and implement algorithms every week. During the lectures we will discuss the last week's homework and material. At the end of each meeting I will present next week's topic and specify the homework.
Examination
To pass the course you are required to complete a minimum of 80 percent of the homework assignments.
Prerequisites
A decent background in topics such as statistics, estimation theory, and stochastic processes is useful.
Mer information
Lennart Svensson
Telephone: 031-772 1777
E-mail: lennart.svensson@chalmers.se
Telephone: 031-772 1777
E-mail: lennart.svensson@chalmers.se
Kurslitteratur
We will use select parts of the two books,
- Branko, R., Arulampalam, S., Gordon, N., (2004). "Beyond the Kalman filter" Artech House, Boston, Mass; London.
- Blackman, S., Popoli, R., (1999). "Design and Analysis of Modern Tracking Systems" Artech House, Norwood, MA.
Föreläsare
Lennart Svensson
E-mail: lennart.svensson@chalmers.se
