ECE 8540 Analysis of Tracking Systems


This course covers topics related to tracking systems, focusing on the filtering methods used to mitigate noise. It is assumed the student has a strong mathematical background and is proficient in MATLAB, C and/or Python programming.

Topics include model fitting and the normal equations, nonlinear soultions, Kalman filter, extended Kalman filter, particle filter, unscented transform, recursive Bayesian estimation, and hidden Markov models. In labwork, several of these ideas are implemented by the student.

Instructor: Adam Hoover

Grader: No grader this semester due to budget crunch

Syllabus

Here is the tentative outline for the semester. Dates and topics are subject to change.

Day Lecture topics (instructor notes) Extra references Lab assigned Lab due
Thursday, 8/19 Introduction to tracking systems and filtering; Line fitting, least squares Zoom A paper on least median of squares fitting to overcome outliers.
Tuesday, 8/24 Normal equations, fitting a circle to a set of data Zoom An excerpt on the normal equations from Numerical Recipes in C. lab 1 - line and model fitting The data for part three of the lab.
Thursday, 8/26 Root finding, nonlinear regression Zoom An online function graphing calculator which can be used to visual the second example in the lecture notes. A discussion of methods used to calculate square roots.
Tuesday, 8/31 Nonlinear model fitting Zoom Some C code that demonstrates fitting a nonlinear model of exponential form. Another C code example that demonstrates fitting a sinusoidal form. lab 2 - nonlinear regression The data files A B and C for fitting. lab 1 due
Thursday, 9/2 Professional writing tools, TeX, vector graphics Zoom A simple example of a TeX file. Another example of a TeX file of the notes for the previous lecture, along with EPS files for the scalar figure and distribution figure. Examples of vector vs raster graphics.
Tuesday, 9/7 Technical writing, content and organization Zoom An example of these writing principles in the business world. From the movie Margin Call about the 2008 financial crisis. lab 2 due
Thursday, 9/9 Writing: graphics and figures Zoom Top 10 worst figures in published research papers. Data distortion, ambiguity and distraction. Anatomy of a graph. Funny examples of selecting models to fit to data. lab 3 - professional writing
Tuesday, 9/14 Filtering, state spaces, state transition and measurement equations Zoom
Thursday, 9/16 Dynamic and measurement noises, optimal balancing Zoom
Tuesday, 9/21 Covariances, matrix notation for filtering Zoom An excerpt from Brookner's book on matrix notation for the KF. lab 3 due
Thursday, 9/23 Kalman filter Zoom An Introduction to the KF by Greg Welch and Gary Bishop. Instructions on using TrackSim to demonstrate the Kalman filter. lab 4 - Kalman filter The data to be used for part 1. Some 2D UWB tracking data that can be used for part 2.
Tuesday, 9/28 2D Kalman filter example Zoom An online demo of a 2D constant velocity model Kalman filter.
Thursday, 9/30 Nonlinear filtering, Jacobians Zoom
Tuesday, 10/5 Extended Kalman filter, radar example Zoom lab 4 due
Thursday, 10/7 EKF example, sinusoid Zoom Instructions on using TrackSim to demonstrate the extended Kalman filter. lab 5 - extended Kalman filter The data to be used for part 1. Some simulated ballistic data for part 2 is forthcoming.
Tuesday, 10/12 no class - Fall break
Thursday, 10/14 Recursive Bayesian estimation Zoom
Tuesday, 10/19 Non-Gaussian state and noise distributions Zoom
Thursday, 10/21 Importance sampling, sequential importance sampling Zoom lab 5 due
Tuesday, 10/26 Particle filter, resampling Zoom A tutorial on particle filtering in the context of signal processing. Another tutorial in the context of mobile robot localization. lab 6 - particle filter The data to be used. Some C code that generated the data, in case you would like to try your own.
Thursday, 10/28 Particle filter demo, coding Zoom Instructions on using TrackSim to demonstrate the particle filter.
Tuesday, 11/2 Introduction to ultra-wideband position tracking Zoom Bill Suski's defense slides
Thursday, 11/4 Noise modeling for UWB indoor position tracking Zoom Salil Banerjee's defense slides
Tuesday, 11/9 Matrix square root, unscented transform A presentation on the unscented particle filter, by Rudolph van der Merwe
Thursday, 11/11 Introduction to hidden Markov models and the Viterbi algorithm Zoom An introduction to HMMs by Raul Ramos. Viterbi algorithm by Didier Gonze. lab 7 - HMM
Tuesday, 11/16 HMM problem solving, example applications. Zoom Examples of HMM applications by Raul Ramos. HMMs for recognizing eating gestures by Yiru Shen. lab 6 due
Thursday, 11/18 Behavior-based robotics Zoom lab 8 - robot simwars. The simulated environment for robot survival wars is available in two versions: C code for Microsoft Visual C and C code for Microsoft Visual Studio 2010.
Tuesday, 11/23 Mobile robot architectures Pictures of different mobile robots explained during lecture. lab 7 due
Thursday, 11/25 no class - Thanksgiving break
Tuesday, 11/30 Robot survival simulations Zoom lab 8 first robot code due
Thursday, 12/2 Robot survival simulations with memory lab 8 second robot code due, report due


ECE 8540 Page / Clemson / ahoover@clemson.edu