ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange
Ask Your Question
0

Building msg in c++

asked 2019-04-09 09:14:52 -0500

THordijk gravatar image

updated 2019-04-09 09:47:28 -0500

gvdhoorn gravatar image

I am trying to construct a message in c++. I am using Ros kinetic, ubuntu 16.04. I have a message that consists of a list of the Point message from geometry_msgs. The message looks as follows:

geometry_msgs/Point[] points

I use this message to transport a list of xy coordinates. The next node uses these coordinates to filter a distance from a pointcloud msg. The part were it goes wrong is when I try to construct the message that the second node publishes. This has to be a list, similair to the one that is sent from the first node. However, this should consist of xyz coordinates. The code I am using (if more is desired I'll add it):

void pixelTo3DPoint(const sensor_msgs::PointCloud2 pCloud, coord_proces::xyz xy_coord)
{
    // get width and height of 2D point cloud data
    int width = pCloud.width;
    int height = pCloud.height;
    int step;
    int n = 0;
    float min_x = 0;
    float min_y = 0;
    float min_z = 0;

    int list_size = xy_coord.points.size();

    float z_List[list_size];
    float y_List[list_size];
    float x_List[list_size];     

    coord_proces::xyz obj_coords;
    geometry_msgs::Point pt;


    for (int i = 0; i < list_size; i++)
    { 
        const int u = static_cast<int>(xy_coord.points[i].x);
        const int v = static_cast<int>(xy_coord.points[i].y);

        // Convert from u (column / width), v (row/height) to position in array
        // where X,Y,Z data starts
        int arrayPosition = v*pCloud.row_step + u*pCloud.point_step;

        // compute position in array where x,y,z data start      
        int arrayPosX = arrayPosition + pCloud.fields[0].offset; // X has an offset of 0
        int arrayPosY = arrayPosition + pCloud.fields[1].offset; // Y has an offset of 4
        int arrayPosZ = arrayPosition + pCloud.fields[2].offset; // Z has an offset of 8

        float X = 0.0;
        float Y = 0.0;
        float Z = 0.0;

        // Copy the data directly from the location in the message
        memcpy(&X, &pCloud.data[arrayPosX], sizeof(float));
        memcpy(&Y, &pCloud.data[arrayPosY], sizeof(float));
        memcpy(&Z, &pCloud.data[arrayPosZ], sizeof(float));

        if(not(isnan(X)) & not(isnan(Y)) & not(isnan(Z)))
        {
            if((0.05 < Z) & (Z < 2.0))
            {
                ROS_INFO("check1");
                ROS_INFO_STREAM(X);
                pt.x = X;
                pt.y = Y;
                pt.z = Z;
            }
        } 

        obj_coords.points[i].insert(0, pt);   
    }

    // select lowest Z
    ROS_INFO("check3");
    point_pub.publish(obj_coords);  
}

The program progresses to check2, but does not reach check3 after constructing the message. I know for a fact that the values in pt are correct. However it won't store them in obj_coords.points. I am guessing this is caused by my attempt to itterate through the list adding points until all relevant xy combinations are combined with their appropriate z.

How would I go about properly storing all coordinates in the message?

edit retag flag offensive close merge delete

Comments

Just an observation:

float X = 0.0;
memcpy(&X, &pCloud.data[arrayPosX], sizeof(float));

What is the difference with X = pCloud.data[arrayPosX]?

gvdhoorn gravatar image gvdhoorn  ( 2019-04-09 09:32:10 -0500 )edit

The program progresses to check2, but does not reach check3 after constructing the message.

there is no check3 in the code you show.

However it won't store them in obj_coords.points.

and how have you determined this?

gvdhoorn gravatar image gvdhoorn  ( 2019-04-09 09:32:30 -0500 )edit

Use && instead of & for boolean logic

bsaund gravatar image bsaund  ( 2019-04-09 09:41:35 -0500 )edit

Posting the specific error you are receiving will help us solve the problem

bsaund gravatar image bsaund  ( 2019-04-09 09:43:04 -0500 )edit

gvdhoorn, I took the code from someone else, I am not very familiar with c++ so I could not tell you the difference. If there is none then I will use your suggestion, as it much cleaner. I have determined that it does not reach check 3 (edited in main question) by running the program in combination with the other nodes. The program crashes between check2 and 3.

bsaund What would be the functional difference between & and &&, that part of the code seems to be working properly.

THordijk gravatar image THordijk  ( 2019-04-09 09:50:23 -0500 )edit

Use && instead of & for boolean logic

this is a good comment, but in the end it doesn't really matter here I believe: the if-clause consists of two predicates. With the boolean<->integer coercion in C++, the clause translates to a binary AND of two integers. The integers will be either 0 or 1. So in the end the entire clause will only be 1 if both subclauses are true == 1.

gvdhoorn gravatar image gvdhoorn  ( 2019-04-09 09:51:58 -0500 )edit

As long as we're "reviewing" this code: the inner loop seems to implement a filter, but the method is called pixelTo3DPoint(..). That seems like a disconnect there.

gvdhoorn gravatar image gvdhoorn  ( 2019-04-09 09:55:02 -0500 )edit

The code takes pixel values x and y and retrieves the corresponding xyz coordinates. What would you call it?

THordijk gravatar image THordijk  ( 2019-04-09 09:57:13 -0500 )edit

So what is this doing in there then?

if((0.05 < Z) & (Z < 2.0))

that looks like it's filtering points based on their Z value.

gvdhoorn gravatar image gvdhoorn  ( 2019-04-09 09:58:29 -0500 )edit

True, I got values far from what was reasonable. I use this to filter a range based on Z values.

THordijk gravatar image THordijk  ( 2019-04-09 10:52:32 -0500 )edit

That would seem to violate the separation of concerns principle.

gvdhoorn gravatar image gvdhoorn  ( 2019-04-09 10:56:41 -0500 )edit

2 Answers

Sort by ยป oldest newest most voted
1

answered 2019-04-09 09:38:20 -0500

gvdhoorn gravatar image

updated 2019-04-09 09:53:20 -0500

coord_proces::xyz obj_coords;
[..]
obj_coords.points[i] = pt;

obj_coords.points is a std::vector (see wiki/msg: Message Description Specification). You cannot just index into a std::vector without first reserving space for the objects. You must either first reserve(..) or you should use push_back(..).

The next node uses these coordinates to filter a distance from a pointcloud msg.

From the code it looks like you're essentially implementing a passthrough filter with a threshold for Z set to min 5cm and max 2m. You might want to take a look at pcl/documentation/Filtering a PointCloud using a PassThrough filter which essentially does the same thing, but in much less lines (and potentially more efficient).


Edit:

The program crashes between check2 and 3.

that would be one symptom of using memory past the end of the vector, which would be another indication that not reserve(..)ing and not using push_back(..) is the most likely cause here.

edit flag offensive delete link more

Comments

This worked thank you!

THordijk gravatar image THordijk  ( 2019-04-09 09:54:53 -0500 )edit

Please note: this is more a general C++ related issue than an actual ROS question. The fact that ROS messages are being used is not that important (in the end it's just a regular C++ std::vector, and that was being used incorrectly).

While we're all here to help, it might be more efficient to post these sort of queries on dedicated C++ fora.

gvdhoorn gravatar image gvdhoorn  ( 2019-04-09 09:56:28 -0500 )edit

Noted. from now i'll go there first.

THordijk gravatar image THordijk  ( 2019-04-09 10:51:13 -0500 )edit
1

answered 2019-04-09 09:40:46 -0500

bsaund gravatar image

Instead of obj_coords.points[i] = pt, use obj_coords.push_back(pt).

obj_coords.points is a vector, and is initialized with size 0, so you are trying to access an element off the end.

edit flag offensive delete link more

Comments

This worked thank you! I did need to keep points in the statement.

THordijk gravatar image THordijk  ( 2019-04-09 09:55:24 -0500 )edit

Question Tools

Stats

Asked: 2019-04-09 09:14:52 -0500

Seen: 880 times

Last updated: Apr 09 '19