Publishing sequences/objects through RTI Conector to ROS2 applications

Question

I am interfacing ROS2 with a native RTI DDS Connector for Python where I am publishing messages in RTI connector and subscribing in ROS2.

I have the following message structure for the message named DetectedObjectList :

int16 id
// An array of objects of another message type
DetectedObject[ ] objects

This is interpreted as unbounded sequences in IDL.

Another message named DetectedObject :

int16 id
string name
int16 x
int16 y

Let's say the topic used for communication is " objects " and the message type is " DetectedObjectList ".

Since the subscriber in ROS2 is subscribing to id of type int16 and objects of type DetectedObject[] , how can I publish an object from RTI connector?

The usual flow in RTI Connector is:

• Get an instance of the output:

  output = connector.getOutput("MyPublisher::MyDataWriter")

• Post an instance:

  output.instance.setNumber("id", 5)

  output.write()

How can I write an object of type DetectedObject instead of setNumber ?

Answer 1

I don't have experience with ROS, but I'll try to help with the DDS/Connector part.

As far as I know in DDS you cannot specify an unbounded array. You can have unbounded Sequences, but not arrays. So, if you are using a type that looks like this:

struct DetectedObject {
  short id;
  string name;
  short x;
  short y;
};


struct MyMessage {
  short id;
  DetectedObject objects[10];
};

or you have an unbounded sequence instead:

struct DetectedObject {
  short id;
  string name;
  short x;
  short y;
};


struct MyMessage {
  short id;
  sequence<DetectedObject> objects;
};

Then your connector code will be something like this:

connector = rti.Connector("MyParticipantLibrary::PubParticipant",
                          filepath + "/ROS.xml")
outputDDS = connector.getOutput("MyPub::MyTopicWriter")

for i in range(1, 500):
    # There are two ways to set values in an instance:

    # 1. Field by Fields:
    outputDDS.instance.setNumber("id", 1)
        #note index, for now, starts from 1. This may change in the future
    outputDDS.instance.setNumber("objects[1].id", 2)
    outputDDS.instance.setString("objects[1].name", "aName")
    outputDDS.instance.setNumber("objects[1].x", 3)
    outputDDS.instance.setNumber("objects[1].y", 4)
    outputDDS.write()

        # OR

    # 2. By first creating a dictionary and then setting it all at once:
    myDict = {'id': 5, 'objects': [{'id': 6, 'name': '', 'x': 7, 'y': 8}]}
    outputDDS.instance.setDictionary(myDict)
    outputDDS.write()
    sleep(2)

Maybe somebody else can contribute more about the ROS <--> DDS mapping when it comes to unbounded arrays.

I hope this help, Gianpiero