Books about robot navigation

[AllanBertelsen]

Hi -
I have been looking for books covering robot navigation. "Sensors for Mobile Robots" looks fine but it's expensive. Also "Mobile Robotics: A Practical Introduction" looks interesting.
There is many books on the market. I bought "Building Robot Drive Trains" and I am very pleased with this book. But it does not cover navigation.
Do you have a favourite?
Can you say: "This book told me everything I needed to know about robot navigation"?
Then tell me the title of the book.

I have been googling, I have been Amazoning and I have looked at http://www.robotbooks.com. I want experienced robot builders advice/opinions on book titles.

 

[dknguyen]

What kind of robot navigation? GPS? localized "GPS"? localized non-RF GPS?Dead-reckoning? Inertial guidance? Obstacle avoidance? Target tracking? Path planning? Landmarking? It's a huge field! I don't think you can find one book on all of it. Narrow your search maybe? You will probably be able to find the books (or more likely university papers) on a specific method.

 

[AllanBertelsen]

Hi dkngyen
You are right. It's a big world. What I'm interested in is answers to a lot of questions.
What sensors are good for what – precision etc?
Are the best practises for Dead-reckoning?
Are there a set of best practises for outdoor positioning?
How do I implement sensor fusion?
And about everything you mention (GPS, Land marking, lasers scaning, rfid, path planning).

For motor control PID is a sort of best practise. There must be a set of best practises in robot navigation.

I have been seeking the internet for some time to try covering this field. But usually a book does a much better job organizing complex stuff.

 

[dknguyen]

A lot of those (mostly the dead-reckoning, landmarking, path planning) are research topics so you could just search online specifically for PDF format papers for those- and books. There's probably also courses about those (indirectly) at universities.

Not sure about sensor fusion (but I'm betting there it requires boatloads of University courses, none of which are actually directly focused on sensor fusion.) Like Kalman filters...it takes a huge amount of math background to understand how they work, let alone implement = more books needed. Although, they do use these in university group projects, but there are team members with different specialities rather than one person.

For sensors...well there are gyros, accelerometers, rangefinders, etc. It's pretty easy to be intuitive (or just Google) to figure out what each can do and what it can't. It's also fairly easy to figure out the advantages and disadvantages of the different approaches to a sensor (ie. for rangefinding you can use laser, sonar, radar. For gyros you have piezo, ring, wine glass, beam, laser, mechanical). Most of the time you are pretty much limited to one method anyways, since the others are way aerospace-grade expensive.

It's all military, aerospace, and research- so look accordingly. Would they sell these types of books on Amazon? Probably not (although I did find and buy one which was a collection of research papers on flapping wing aerodynamics, many of which I had already downloaded off of the internet. It is SOOOO way over my head and probably always will be, since I am not a mechanical engineer..s o if anyone wants to buy it off me they can! Mint condition!). The problem with robot navigation is that it is not mature- it is research. Even worse, it's an application (which involves many fields) rather than a field. Chances are that any book you find is going to be so technical that it is way over your head(although I am not sure how versed you are). Have you ever heard of a Kalmnan filter? It's used in sensor fusion I believe. It takes a huge amount of math to understand, let alone implement. These are not things you can just gloss over. As such, no one writes books that combine all of this together. As a result of this, it's probably not even referred to as robot navigation (except in research papers and journals).

For example, path planning is an application- an immature, research-level application (with the exception of some application aerospace/military). As such, you probably can't find a book called path planning. The book would probably be some specific generic mathematical/AI method that can be used for as a specific method of path planning. The title is probably a word used in AI to describe algorithms for data collection, interpretation, and optmization.

If you still are looking for a book:
-Don't search for robot navigation, it's so general you'll probably never find anything (particularily on the level you want)
-You'll need to get much more specific
-You'll need much more than one book
-Don't search for a book with robots on the label (robots tends to imply hobbyist on a book a lot of the time. This is not hobbyist stuff)

If I were you, I would not look for books so much as research journals and graduate research papers. Of course, then you will need to get things like advanced statistics and math textbooks to figure the stuff out (but that would probably happen anyways even with a book, unless it was a meter thick.) Normally, I just read those papers to get a conceptual feel. Then I simplify it to something I can actually pull off on my own.

When these things are used in real systems, you have a bunch of engineers each specialized in their particular field. That's how they can use all this stuff. If you are just one person, you can just get creative. Use the sensors you can afford and use more simplistic methods, like weighted averages with guesstimated weightings. Intelligently comparing different sensor readings (using your intuition to come up with the stuff and the field testing it). Like if the readings from two different types of sensors don't "overlap" enough, prioritize one over the other, or reject them altogether. Things like that since you probably don't have the time, money, or ability to learn and implement all of these algorithms and expensive sensor technologies.

 

[AllanBertelsen]

My info junk box

Thank you for the views and advise dknguyen. I found a lot of PDF's on sensor fusion and among these also Kalman filters and a lot of material describing all the subjects I seek information about. But it is like I'm emptying a junk box and claiming that I got a robot kit. Nearly all the parts necessary are there but a lot of nuts and bolts are missing. In this analogy, the book would be a real robot kit. Not just a junk box.
But if this is the state of things today I got to accept it. A field where laboratories are still seeking answers in the dark?!

At least I want to share some links to untraditional solutions for the hobbyist that I found:
The newsletter of the Seattle Robotics Society is something you got to read if you at robot hobbyist.

An example is an article as: Outdoor Navigation using Polarized Light
Another is: 'A Real-time Laser Range Finding Vision System'

Then I found Alex Brown's Robotics Page at **broken link removed** He shows the use of gyros an compass as well as using a laptop for robot "brain".

Another hobbyist page with a laser rangefinder is https://jormungand.net/projects/ And another very good laser rangefinder at **broken link removed**

Then there is all the heavy stuff. Universities all over the world publish PDFs for free.
I found a lot at Nanyang Technological University Singapore. But also Carnegie Mellon University
Pittsburgh has a lot to offer. **broken link removed** is just one example. And there is a lot of semilar sites loaded with articles. Sidney Australia. University of British Columbia Vancouver not to forget.

And Google also really has something to offer. Google runs a beta on a new service called Goggle Scholar. Try for example this link https://scholar.google.com/scholar?num=20&hl=da&lr=&q=robot+localization

So this is some of the parts in my info junk box. Hope someone can use some of the bits for somthing. Seems to that we have to write that book ourselves.