Work at SourceForge, help us to make it a better place! We have an immediate need for a Support Technician in our San Francisco or Denver office.

## mrpt-help — Help on MRPT compilation and usage

You can subscribe to this list here.

 2008 2009 2010 2011 2013 2014 Jan Feb Mar Apr May Jun Jul Aug Sep (6) Oct (3) Nov Dec (2) Jan (7) Feb (12) Mar (20) Apr (12) May (16) Jun (15) Jul (12) Aug (16) Sep (17) Oct (51) Nov (22) Dec (5) Jan (20) Feb (44) Mar (37) Apr (41) May (63) Jun (64) Jul (74) Aug (62) Sep (46) Oct (62) Nov (65) Dec (68) Jan (113) Feb (56) Mar (111) Apr (75) May (76) Jun (120) Jul (65) Aug (20) Sep (76) Oct (3) Nov Dec Jan (1) Feb Mar (4) Apr (3) May (1) Jun Jul (2) Aug Sep Oct Nov Dec Jan Feb Mar Apr (2) May (3) Jun Jul Aug (1) Sep Oct Nov Dec

Showing results of 1587

<< < 1 2 3 4 5 .. 64 > >> (Page 3 of 64)
 [Mrpt-help] [mrpt.org RSS update] Forums-users: Getting data by using doProcess() function of Hokuyo sensor From: noreply - 2011-09-12 07:40:16 MRPT.ORG RSS updateNew entry: Forums-users: Getting data by using doProcess() function of Hokuyo sensor
Original item: http://www.mrpt.org/node/1499

Hi,

I tried to acquire data from Hokuyo sensor by following code.

<code>

do{
cin>>c;
sensorHokuyo->doProcess();
sensorHokuyo->getObservations(lstObjs);
if(c[0]=='o'){
CObservationPtr obs = CObservationPtr(lstObjs.begin()->second);
CSensoryFrame SF;
SF.insert(obs);
out_file<<SF;
}
sleep(20);
}while(true);


 [Mrpt-help] [mrpt.org RSS update] I don't have a copy of From: noreply - 2011-09-10 09:30:10 MRPT.ORG RSS updateNew entry: I don't have a copy of
Original item: http://www.mrpt.org/node/1476#comment-682

In reply to Need some more explanation for variable of COccupancyGridMap:

I don't have a copy of Probabilistic Robotics at hand, but according to this Thrun's paper:

Learning occupancy grid maps with forward sensor models (PDF)

;

I guess you intend to use the "forward model" to build the map. That approach is not implemented in MRPT so far, the only method is the "inverse model" approach, which as Thrun writes, assumes independence between all cells. However, it is a quite good solution in general, fastest than the "forward method" and for precise laser scanners there is not much difference between both of them.

To update the map from a Hokuyo (or any other scanner) scan, please use the method: mrpt::slam::COccupancyGridMap2D::insertObservation(), documented in the virtual base class CMetricMap;. For example:

COccupancyGridMap2D  gridmap; gridmap.insertionOptions.... = ...; // Set grid options...  CObservation2DRangeScan  scan;  CPose3D  pose3D(CPose2(x,y,phi_rad)); gridMap.insertObservation( &scan, &pose3D );


 [Mrpt-help] [mrpt.org RSS update] Thank you very much for the From: noreply - 2011-09-10 09:15:10 MRPT.ORG RSS updateNew entry: Thank you very much for the
Original item: http://www.mrpt.org/node/1476#comment-681

In reply to Need some more explanation for variable of COccupancyGridMap:

Thank you very much for the answers :)

Actucally, I

One more thing, how can I find Inverse Measurement Model for my own sensor(Hokuyo)? Do you have a class that can find the sensor model and use that parameters in your COccupancyGridMap class?

In the "Probabilistic Robotics", they suggested some like generating samples from the measurement model or approximating the inverse using a supervised learning algorithm such as logistic regression or a neural network.


 [Mrpt-help] [mrpt.org RSS update] Just for the records: The From: noreply - 2011-09-09 23:25:10 MRPT.ORG RSS updateNew entry: Just for the records: The
Original item: http://www.mrpt.org/node/453#comment-680

In reply to Please have a look at my driver for the Ibeo Lux scanner:

Just for the records: The contributed code was long ago integrated into MRPT in the class:  mrpt::hwdrivers::CIbeoLuxETH

; 
 [Mrpt-help] [mrpt.org RSS update] Hi, I have read From: noreply - 2011-09-09 22:50:10 MRPT.ORG RSS updateNew entry: Hi, I have read
Original item: http://www.mrpt.org/node/1476#comment-679

In reply to Need some more explanation for variable of COccupancyGridMap:

Hi,

I have read "Probabilistic Robotics", there are 2 methods mentioned 1.) using inverse_sensor_model 2.)Forword model which one do you use? and I think this value must be corresponding to sensor model I use, right?

The inverse sensor model is the one applicable here while updating a grid map from a laser scan. The "forward model", if I recall it right, was the name for the sensor model that makes predictions. That is what you select in the likelihoodOptions struct in MRPT, but is a different story...

"mapAltitude" (z-axis) = does this mean the height of sensor I attach to the robot?

Yes, but it only if useMapAltitude is set to true.

"maxDistanceInsertion" = the max range of sensor I use?

Yes and no. It can be equal to the sensor max range but you can choose to limit its effects while updating the grid to something shorter for some reason.

"maxOccupancyUpdateCertainty"

This must be between 0.5 and 0.999..., and is the probability of a cell being occupied in the inverse sensor model, which is used in the log-odds Bayesian update of the grid.

In general, the params are more or less described here: http://reference.mrpt.org/svn/classmrpt_1_1slam_1_1_c_occupancy_grid_map2_d_1_1_t_insertion_options.html

;

but it's also true that the documentation page for; occupancy grid maps has been in the TO DO for too long, but I hope the class is understadable in the meanwhile.

Cheers,

JL


 [Mrpt-help] [mrpt.org RSS update] I am figuring this out,too. From: noreply - 2011-09-09 13:35:10 MRPT.ORG RSS updateNew entry: I am figuring this out,too.
Original item: http://www.mrpt.org/node/736#comment-678

In reply to Build a map without SLAM:

I am figuring this out,too. Just started with MRPT. Don't know there is complete application or example provided or not.
However, I think for my consideration, we should use Occupancy grid map class and build the map with the exact pose.


 [Mrpt-help] [mrpt.org RSS update] Forums-users: Need some more explanation for variable of COccupancyGridMap From: noreply - 2011-09-09 11:15:10 MRPT.ORG RSS updateNew entry: Forums-users: Need some more explanation for variable of COccupancyGridMap
Original item: http://www.mrpt.org/node/1476

Hi,

At the present, I am trying to create map in my room using Pioneer P3DX with Hokuyo sensor. I don't want to use SLAM for now (since I still don't understand how it work). Thus, I only try to put robot on the exact pose in the room and acquire the data. And I will use COccupancyGridMap to build map from these data.


 [Mrpt-help] [mrpt.org RSS update] Forums-users: Problem building Mrpt From: noreply - 2011-09-08 11:55:10 MRPT.ORG RSS updateNew entry: Forums-users: Problem building Mrpt
Original item: http://www.mrpt.org/node/1421

Hello,
I've been trying to build MRPT for a while now, finally succeeded, but I get some errors while trying to build the test by inputting the "make test" command.
Here's the output of my terminal :


 [Mrpt-help] [mrpt.org RSS update] Forums-users: ICP3D_Method_Classic() lacks RANSAC implementation??? From: noreply - 2011-09-08 11:55:10 MRPT.ORG RSS updateNew entry: Forums-users: ICP3D_Method_Classic() lacks RANSAC implementation???
Original item: http://www.mrpt.org/node/1403

I've dug into the source code related to the ICP algorithm. When looking in CICP.cpp (http://code.google.com/p/mrpt/source/browse/trunk/libs/slam/src/slam/CIC...;) I noticed that the function ICP3D_Method_Classic() lacks a RANSAC implementation of any kind, while the 2D versions of ICP do have this implemented. I was wondering why there is no RANSAC implementation for ICP for the 3D case, but is for the 2D case.


 [Mrpt-help] [mrpt.org RSS update] Thanks a lot for your From: noreply - 2011-09-08 11:40:13 MRPT.ORG RSS updateNew entry: Thanks a lot for your
Original item: http://www.mrpt.org/node/1469#comment-675

In reply to Weight equation in PF in bayesianTracking example.:

Thanks a lot for your clarification!! Your library is very cool and easy to use for me :)
Great to hear I can help some.


 [Mrpt-help] [mrpt.org RSS update] Btw, I still don't get it. From: noreply - 2011-09-08 11:40:13 MRPT.ORG RSS updateNew entry: Btw, I still don't get it.
Original item: http://www.mrpt.org/node/1467#comment-677

In reply to Need some explanation of bayesianTracking Example:

Btw, I still don't get it. What is the meaning of the value of ESS?
if that time it gets 1.0, what does it mean? if less than 1.0, it's not good?

>In MRPT it includes a little extra logic to firstly normalize the weights which are stored as log-weights.
Do you mean we don't have to make extra code to normalize the weight at all? MRPT've just done it for us?


 [Mrpt-help] [mrpt.org RSS update] Forums-users: comprar viagra para mujeres From: noreply - 2011-09-08 06:45:10 MRPT.ORG RSS updateNew entry: Forums-users: comprar viagra para mujeres
Original item: http://www.mrpt.org/node/1474


 [Mrpt-help] [mrpt.org RSS update] Forums-users: RawlogViewer stop working.. From: noreply - 2011-09-08 04:25:11 MRPT.ORG RSS updateNew entry: Forums-users: RawlogViewer stop working..
Original item: http://www.mrpt.org/node/1473

hi all,

I'm just downloaded the MRPT and do my project.but when i try to load the rawlog dataset into rawlogviewer, it will stop immediately and shown this error: "an unhandled win32 exception occurred in rawlogviewer.exe [4360]. I'm using windows 7 pro currently.

Any hints how to solve this??


 [Mrpt-help] [mrpt.org RSS update] MRPT blog: 08/SEP/2011: Mathjax in MRPT forums From: noreply - 2011-09-07 22:30:11 MRPT.ORG RSS updateNew entry: MRPT blog: 08/SEP/2011: Mathjax in MRPT forums
Original item: http://www.mrpt.org/node/1472

From today, anyone can write Latex equations in the MRPT forums, thanks to the Mathjax library.

It's as easy as writing your Latex formulas enclosed within (remove the spaces between the backslash and the square brackets):

\ [  Your Latex code \ ]

For example:

\ [  \int_0^x \sin^2(z) dz = \frac{\partial F(x)}{\partial x} \ ]

renders as:


 [Mrpt-help] [mrpt.org RSS update] No, you don't need to do so. From: noreply - 2011-09-07 22:05:11 MRPT.ORG RSS updateNew entry: No, you don't need to do so.
Original item: http://www.mrpt.org/node/1469#comment-674

In reply to Weight equation in PF in bayesianTracking example.:

No, you don't need to do so. On line 595 weights are added because they are log-weights, thus in fact we are doing the usual PF weight update for the "standard" proposal distribution, which is a multiplication:

After that, weights are internally renormalized, the user of the MRPT API doesn't have to worry about that.

> But, if look at line 649 in getMean function, weights are always reset to 0 here.

Thanks for noticing it!!
That was a stupid copy&paste bug which has been there for years without noticing. Indeed: no, you don't have to reset the weights (just committed a fix to SVN).

Cheers,
JL


 [Mrpt-help] [mrpt.org RSS update] Hi there, >In line 604 in From: noreply - 2011-09-07 22:00:10 MRPT.ORG RSS updateNew entry: Hi there, >In line 604 in
Original item: http://www.mrpt.org/node/1467#comment-673

In reply to Need some explanation of bayesianTracking Example:

Hi there,

>In line 604 in initializeParticles function:
>I wonder why you decided the value "-VEHICLE_INITIAL_V" for vx and "0" for vy?
> (Just clarify whether there is implicit meaning.)
> I think whatever here is fine for the initial since at last they will converse to real values.

They can be set to anything, more or less close to the expected values. At t=0, the simulator has a +VEHICLE_INITIAL_V for vx (so we are putting things hard to the estimators there!) and a vy=0 (so we are leaving that one easier to track in the first step).

> What is the meaning of ESS (Estimated Sample Size)?

It is a well known statistics in particle filter literature:

In MRPT it includes a little extra logic to firstly normalize the weights which are stored as log-weights.

JL


 [Mrpt-help] [mrpt.org RSS update] Perhaps MRPT was compiled to From: noreply - 2011-09-07 21:35:10 MRPT.ORG RSS updateNew entry: Perhaps MRPT was compiled to
Original item: http://www.mrpt.org/node/1471#comment-672

Perhaps MRPT was compiled to use SSE* optimizations and the processor doesn't support them?

Try two things: debugging with gdb ("gdb PROGRAM", "run", on crash, type "bt" to see the stack) or just rebuild MRPT in the laptop checking the DISABLE_SSE option in CMake, which is hidden as "advanced" by default.

Hope that helps,
JL


 [Mrpt-help] [mrpt.org RSS update] Forums-users: FATAL error From: noreply - 2011-09-07 19:50:11 MRPT.ORG RSS updateNew entry: Forums-users: FATAL error
Original item: http://www.mrpt.org/node/1471

Hi,


 [Mrpt-help] [mrpt.org RSS update] Forums-users: Need some explanation of bayesianTracking Example From: noreply - 2011-09-07 15:15:15 MRPT.ORG RSS updateNew entry: Forums-users: Need some explanation of bayesianTracking Example
Original item: http://www.mrpt.org/node/1467

Hi,

Here, I am try to learn Particle Filter from bayesianTracking example. However, I am confused with some values and function used in the code as following:

In line 604 in initializeParticles function:
I wonder why you decided the value "-VEHICLE_INITIAL_V" for vx and "0" for vy? (Just clarify whether there is implicit meaning.)
I think whatever here is fine for the initial since at last they will converse to real values.


 [Mrpt-help] [mrpt.org RSS update] Forums-users: Weight equation in PF in bayesianTracking example. From: noreply - 2011-09-07 15:15:13 MRPT.ORG RSS updateNew entry: Forums-users: Weight equation in PF in bayesianTracking example.
Original item: http://www.mrpt.org/node/1469

Hi,

I am looking at bayesianTracking example and wonder whether we have to reset weight to 0 every loop or not?
>From the example code, line 595, weights are always plus to the original value. But, if look at line 649 in getMean function, weights are always reset to 0 here. Therefore, what is right? reset to 0? or not reset?
Because if it has to reset every loop why don't you just assign = in line 595 instead of += at the first place?

parnurzeal


 [Mrpt-help] [mrpt.org RSS update] Apparently - reading the From: noreply - 2011-09-07 15:00:13 MRPT.ORG RSS updateNew entry: Apparently - reading the
Original item: http://www.mrpt.org/node/1456#comment-671

In reply to Is this a book of mrpt library?:

Apparently - reading the stuff here http://www.chrisrand.com/blog/index.php/2010/02/27/odd-tale-alphascript-...

;

you can tell Amazon about breach of copyright and the listing will be removed


 [Mrpt-help] [mrpt.org RSS update] Hi JL, thanks for your help. From: noreply - 2011-09-07 12:55:13 MRPT.ORG RSS updateNew entry: Hi JL, thanks for your help.
Original item: http://www.mrpt.org/node/1457#comment-670

In reply to Cannot run HMT-SLAM:

Hi JL,

thanks for your help. However I still cannot run hmt-slam which is very unfortunate as it (as far as I know) is the only hybrid slam approach with available code.

Thank you though!

MH


 [Mrpt-help] [mrpt.org RSS update] Forums-users: cheap generic Linezolid canada From: noreply - 2011-09-07 06:15:09 MRPT.ORG RSS updateNew entry: Forums-users: cheap generic Linezolid canada
Original item: http://www.mrpt.org/node/1466


 [Mrpt-help] [mrpt.org RSS update] Forums-users: cheap Tequin canadian sales - american pharmacies for generic Tequin 400mg online kopen From: noreply - 2011-09-07 05:15:10 MRPT.ORG RSS updateNew entry: Forums-users: cheap Tequin canadian sales - american pharmacies for generic Tequin 400mg online kopen
Original item: http://www.mrpt.org/node/1465


 [Mrpt-help] [mrpt.org RSS update] Forums-users: online order Rheumatrex australia over the counter - canadian pharmacy Rheumatrex 2.5 mg online fast delivery From: noreply - 2011-09-07 03:40:11 MRPT.ORG RSS updateNew entry: Forums-users: online order Rheumatrex australia over the counter - canadian pharmacy Rheumatrex 2.5 mg online fast delivery
Original item: http://www.mrpt.org/node/1464