I am currently working on the "likelihood field" sensor model (aka "endpoint model") by Thrun et al. (book: Probabilistic Robotics, Thrun et al., p.169-174) In this context I have two questions. According to the authors, the probability of a sensor measurement $z_{t}^{k}$ can be described by a weighted set consisting of $p_{hit}$, $p_{rand}$ and $p_{max}$ (mixing weights $z_{hit}$, $z_{rand}$, $z_{max}$):

Probabilistic Robotics, Thrun et al., p.171

In the following, the authors describe that the implementation of the likelihood field algorithm (see below) does not take $z_{max}$ measurements into account.

Question 1: If $z_{max}$ measurements are rejected, what is the intention of the probability $p_{max}$ in equation (6.34)?

My second question refers to line 8 of the following algorithm:

Probabilistic Robotics, TRHUN et al., p.172

Note errata : replace "$z_{random}$" by "$z_{rand}$" (http://probabilistic-robotics.informatik.uni-freiburg.de/errata.html)

What is the meaning of the second term $\frac{zrand}{zmax}$? Does $z_{rand}$ describe the weight factor of the uniform distribution $p_{rand} = \frac{1}{z_{max}}$? In this context, $z_{max}$ describes the maximum range of the laser scanner and not a weight factor...

Thanks for your help.



Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.