Customer Reviews

Tracking and Kalman Filtering Made Easy

5 star:		(5)
4 star:		(1)
3 star:		(0)
2 star:		(0)
1 star:		(1)

 

 

Brookner builds up from the simplest case (a target moving at constant speed and constant heading) and gradually builds these simple equations into the simplest-case Kalman filter. And this is just in the first chapter! His approach is intuitive; he is rigorous enough, but this book is probably more for engineers than for mathematicians. He develops everything quite logically; in the very earliest going, normal algebraic equations are used and only when it makes sense does he switch to linear algebra, explaining what its advantages are. Great book! Even though the job disappeared, I kept my copy (despite the price!)

As a software developer, I was looking for help in developing an algorithm for tracking targets. This book fits the bill perfectly. The author starts with very basic examples, and shows step-by-step how to increase the complexity to work for a myriad of problems. He also shows how many of the filters used for tracking problems are related (mostly just special cases of a very simple algorithm), and when one should be used instead of another (or in some cases why and how to use multiple filters together). The only downside is that there are multiple typos; perhaps the publisher can fix these in the next edition.

The author tries to make the subject easier than it can be made. He does mention general concepts but does not mention under what circumstances these hold. The danger comes when you try to implement it. Of course you get an answer but as the author puts it himself: "surprise!" the answer is wrong because you were not told about the validity of your assumptions. As an example take the issue of the discrete noise covariance matrix i.e. Q. The author gives a simple recipe for constructing this matrix. But he does not tell you that it is only valid for small step size. If you try larger time steps you are in big trouble. The book contains occasional insightful remarks. Also the same filter has several different names in the same paragraph, which can cause confusion.

This is a graduate level book, but contains everything an undergraduate, engineer, or scientist needs as a reference on the subject. If I was looking for a professor to study under, they would be required to list this book in their syllabus. A software development engineer will find the basis for developing algorithms, or maintaining production software. 16 Pages of Symbol definitions, 37 Pages of Solutions, 8 pages of References.

I have been desperately trying to find a resource that does precisely what the title of this book says - making the Kalman filter easy. It is not difficult to find papers in the literature with this stated aim - providing the intuition behind the matrix algebra but failing to connect the simple intuition with the actual matrix representations. This book does this.

With the intuition developed, the reader should be prepared to go on to more detailed and complete mathematical references on the subject.

This is a great place to start in your study of tracking. The book described things in a simple manner from the ground up in a progressive manner. The book does use unusual notation, but explains it better than any other book I have ever seen. Every single variable is defined in the back of the book, which makes quick reference to the book very easy compared to many books. The book is somewhat light on theory, and is a little out of date, but it is an excellent place to start.

Brookner makes a difficult topic very easy to understand without watering down content. I highly recommend this book to anyone filtering position estimates!