Re: Bayesian Inference Engine

On Sun, 10 Sep 2006 05:55:56 GMT, Michael Olea <oleaj@xxxxxxxxxxxxx>
wrote:

HMSBeagle wrote:

On 9 Jun 2006 04:38:24 -0700, "JGCASEY" <jgkjcasey@xxxxxxxxxxxx>
wrote:
The winner of the DARPA grand Challenge is a Bayesian
Inference Engine.

I dont know who wrote this first. I will go ahead and attribute it to
JGC.

Anyways, its false. The DARPA vehicle that was the WINNER did not use
bayes methods. It used a coupling algorithm between a distance radar
and a visual camera. The basic motto behind the winning car was "the
flat road is this particular color, so drive towards that color
more."

The second-place vehicle, on the other hand, used bayesian methods to
determine the location of obstacles. It was not an "Inference Engine"
at all. It was more like a "cloud" algorithm, where each guess in
the "cloud" of guesses was updated as sensory information came in. The
final determination of the location of the obstacle was obtained by
finding the place in front of the vehicle with the greatest density of
guesses.

What is it that draws idiots to C.A.P. like moths to a flame?

Google: Thrun Bayesian Darpa

The truth is out there.

-- Michael

We will see who the "idiot" is now won't we?

I was only incorrect in saying that the second-place vehicle used a
cloud of guesses for obstacles. That was wrong. It was the winning
vehicle that used them (Stanley). Outside of that, everything else
in my post above is true. To demonstrate, I will quote the pdf
found here:
http://robots.stanford.edu/papers/thrun.stanley05.pdf

This pdf was written by Sebastian Thrun himself. First and foremost,
the word "Bayesian" appears NOWHERE in the pdf. Absolutely nowhere.
The word "bayes" does not appear anywhere in the pdf. The phrase
"inference engine" appears nowhere in the pdf - not even once. So the
claim that "The winner of the DARPA grand challenge was a Bayesian
Inference Engine" is already false.

I will now quote copiously from the pdf itself.

[pdf]
"Obstacle detection on laser point clouds can be formulated as a
classification problem, assigning to each 2-D location in a surface
grid one of three possible values: occupied, free, and unknown.
A location is occupied by an obstacle if we can findnd two nearby
points whose vertical distance exceeds a critical vertical distance
DELTA."
[end pdf]

The laser point clouds are then updated probabilistically as time
progresses, just exactly as I had said in MY post above yours.

[pdf]
"For example, suppose we observe a new measurement for a cell which
was previously observed. Then one or more of three cases will be
true:
1. etc etc
2. etc etc
3. The third case is equivalent to the second, but with a refinement
of the upper value. A new measurement may simultaneously refine the
lower and the upper bounds."
[end pdf]

Notice this describes the REFINEMENT of points in the cloud over time
as *NEW MEASUREMENTS" come in, just exactly as I had described in my
post.

The part of my post that was completely correct was that Stanley
couples the COLOR of flat road with laser sensor data that is
confirming that it is flat. The car then drives towards this color,
assuming that it will not change dramatically over time. Dr. Thrun
himself admits that he did not invent this methodology. I quote Thrun
in his own words:

[pdf]
"To find the road, the vision module classifies images into drivable
and non-drivable regions."

....we already have such drivability information from the laser in the
near range. All that is required from the vision routine is to extend
the reach of the laser analysis.
..
Stanley finds drivable surfaces by projecting drivable area from the
laser analysis into the camera image.
..
The learning algorithm maintains a mixture of Gaussians that model the
color of drivable terrain."
[end pdf]

I repeat for emphasis *THE COLOR OF DRIVABLE TERRIAIN*. This is
exactly what I said in my post, albeit with different wording.
Therefore my original post was correct. I will not continue to
respond to posts in the future that are written by people calling me
names.
.