robzz's profile - activity

Or, point me to the location in that 1700 line file that you're referring to :)

@RickArmstrong Hm I don't see any comments about rotating a vector by a quaternion there. I posted an answer here with some code that seems to work for the few cases that I've tried, using some of those functions. Would you mind taking a look?

Okay, based on @tfoote's response and this answer to a related question, I think this does the job:

# rotate vector v1 by quaternion q1 
def qv_mult(q1, v1):
    # comment this out if v1 doesn't need to be a unit vector
    v1 = tf.transformations.unit_vector(v1)
    q2 = list(v1)
    q2.append(0.0)
    return tf.transformations.quaternion_multiply(
        tf.transformations.quaternion_multiply(q1, q2), 
        tf.transformations.quaternion_conjugate(q1)
    )[:3]

Do I need to do matrix multiplications outside of those functions to accomplish this? If so, which? Better yet: where can I find examples of doing this? I've been search online for such an example using python tf, but with no luck.

Right, so which of those transformation rotates a vector by a quaternion, if any?

Right, so which of those transformation rotates a vector by a quaternion, if any?

What is the python tf API to rotate a vector by a quaternion to get another vector?

I would like to mess around with the ROS navigation stack with a custom robot that I've built. I'm trying to figure out if it can be used to navigate in a previously unknown environment. The tutorial for setting up a robot to use the nav stack states that a map is not required. But this page on map building says that it does requires a static map. Aren't these statements contradictory? Which is true?

I'm trying to calibrate my webcam to remove distortions. I run

$ rosrun usb_cam usb_cam_node

Then this error and warning message is printed out:

[ERROR] [1384614141.059240716]: Unable to open camera calibration file [/home/robz/.ros/camera_info/head_camera.yaml]
[ WARN] [1384614141.060021982]: Camera calibration file /home/robz/.ros/camera_info/head_camera.yaml not found.

Not sure what that means. I wouldn't expect a calibration file to exist for a camera that I haven't been able to calibrate yet...

In any case, the node is still able to run despite the error and warning message. It publishes to a topic, and I can see it with image_view just fine. So then I run

$ ROS_NAMESPACE=usb_cam rosrun image_proc image_proc

Then, this error message is printed out:

[ERROR] [1384614331.794465524]: Rectified topic '/usb_cam/image_rect_color' requested but camera publishing '/usb_cam/camera_info' is uncalibrated

This is very confusing for me. I was under the impression that the image_proc node does the calibration. Why does it require a calibrated camera in order to run?

I have two topics publishing the std_msgs/Float32 message. Is there a way I can plot the data in one topic against the other?

I'm designing a node in my system that needs to gather data from two topics (A and B), process it, and publish the results to another topic. I've considered a few implementation alternatives considering migrating to services, using fixed-rate loops, and retaining the data that a topic handler last received. Here are a few alternatives:

1. Create a handler for each topic. In a loop running at a fixed rate, publish the results using the data that was last received for each topic.
2. Create a handler for each topic. In handler A, publish results using the data from received in that handler with the data that was last received from handler B.
3. Migrate topic A to be a getter service instead. In the handler for topic B, publish results using the data received in that handler and the data received from a call to the new service.

Which method is best? Is there a better way to do this?

I've written a node in python to read a binary image (of type sensor_msgs/Image) being published on a topic. I got confused when trying to access the data in the callback.

print img

produces

header: 
  seq: 35232
  stamp: 
    secs: 0
    nsecs: 0
  frame_id: ''
height: 40
width: 127
encoding: 8UC3
is_bigendian: 0
step: 381
data: [0, 0, 0, 0, ...]

I shortened the data array with an ellipsis.

What's weird are the types of the data.

print type(img.data), type(img.data[0])

produces

<type 'str'> <type 'str'>

When I try to display the data with "print img.data" I get a bunch of junk characters. When I try to index into the first element, I get an empty string.

This confuses me because the description of the topic type Image specifies that the img.data field is an array of bytes. Why am I seeing a string here?

If I have a node which listens to multiple topics & has a callback for each topic, should I need locks/mutexes on the resources that are changed by more than one callback? My question stems from whether or not callback functions are executed in new threads--could code be concurrently executing in two different callback functions at once?

I want to stream live data from my robot to the internet.

Right now I have a ROS node communicating with a webcam and publishing Image/CompressedImage data. The robot needs this data to follow lines/recognize obstacles. I'm trying to figure out how to have the data be sent to show up in a browser in realtime video. I'm not sure on the best way to go about this.

I could have the computer running ROS also be running a server, copying jpeg/png CompressedImage data to a file, then have a client connect to that server to periodically poll for the image and update its page to simulate video..

Is there a more common/better approach?