fsteinhardt's profile - activity

Does roswtf show any problems?

If you are using move_base the result is output to the topic cmd_vel (arrow on the bottom of the image on the move_base webpage ) as a geometry_msgs/Twist message .

I haven't used a car-like kinematic myself, but I just found this page http://wiki.ros.org/teb_local_planner... with more details for the teb planner and ackermann kinematics.

The sections 2.2.1 and 2.2.2. could interest you.

Either cartographer really needs the acceleration values or just has the covariance/value input problem. You could file a cartographer issue and ask, maybe they can follow the -1 covariance convention.

Try setting the first element of the covariance to -1 for the unused measurement fields.

# If you have no estimate for one of the data elements (e.g. your IMU doesn't produce an orientation 
#  estimate), please set element 0 of the associated covariance matrix to -1

(from the IMU message description http://docs.ros.org/api/sensor_msgs/html/msg/Imu.html )

But I don't know if this helps with your cartographer problem, I haven't used it yet.

Has your user the rights to use the serial port on the laptop? Try sudo chmod 666 /dev/ttyS0 (or whatever your serial port is)

You want a python serial port library? something like https://pythonhosted.org/pyserial/ ?

If your code is already running the serial part you are missing the ROS communication part. Have you seen the ROS Python tutorials? http://wiki.ros.org/ROS/Tutorials/Wri... ?

Have you tried to view the serial data with a terminal program? (or just cat /dev/ttyS0 , just to see if the scanner are sending data? ) Configure the Scanners to send data continuously for this test.

Do your ports support RS-422 and 500k baud? Can your S300 send data? The cheaper ones can't.

Hi,

you could try using this package to estimate odometry data from laserscans. http://wiki.ros.org/laser_scan_matcher

Do you really want a SLAM system or do you only want the robot pose?

http://wiki.ros.org/ar_pose and a camera might be enough to get the position and orientation.

There seems to be a typo in your include:
try MultiArrayDimension.h , you typed a 't' in Dimension.

Hi,

you could try the global_localization service of amcl, this spreads the particles all over the map. Then you have to drive around and maybe you get the correct position after the particles converged.

I don't know a package that can do this, but maybe the driver can already invert the laserscan, which scanner do you use? For example the http://wiki.ros.org/cob_sick_s300 driver has a parameter inverted. Or it could be enough to turn the tf laserscanner frame upside down.

 rostopic info /topicname

prints the publishers and subscribers of this topic.

Can you display polygons?

If not you could be having this problem: https://github.com/ros-visualization/...

Your Ubuntu (i guess you are using Ubuntu) package management system seems to be in a partially broken state. This is a Linux/Ubuntu question, not a ROS question. You could search or ask here: http://askubuntu.com or try a completely new Ubuntu installation.

or http://wiki.ros.org/serial for C++

If you can't fix this problem on the software side (I don't know how), you could use a HDMI dummy plug.

It all depends.

Hardware: Laserscanners can be connected over a serial connection or USB or Ethernet, depending on the type of the scanner, IMUs are often connected over Serial or USB, low level sensors like switches or wheel encoders are often connected to a microcontroller which is connected to the main PC.

Software: You need ROS-drivers for the Sensors, for some laserscanners and IMUs you can find ROS packages, or you can program them yourself. Some motorcontrollers have inputs for wheel-encoders.

And now for something completely different: A IMU alone is not able to measure a position in space, and to assemble a 3D scan from a flying not really known position will not be easy. Why don't you try to generate a 3D scan from a fixed position first.

Edit:

for a non flying test: I meant keep the position fixed, but change the orientation of the scanner (tilt it or rotate it), then you don't start with both problems of measuring the orientation and the position at the same time.

Do you know laser_assembler ? This could start you on the software side.

How are these sensors connected to your computer? What kind of robot are you using?

I don't think you can just connect your browser to a websocket server, you need a program that talks the websocket protocol. The server doesn't serve a http page, but wants to upgrade the connection to ws. A Javascript program running in a browser could be used to connect using websocket, see http://wiki.ros.org/roslibjs/Tutorial...

Edit: There is no ROS webbrowser, you can use a standard webbrowser. The Websocket Server is running under ROS. From the roslibjs Tutorial page: Running the Example starts with starting the Server and publishing messages, but the last 5 lines say: You can open the html file in a local webbrowser (open file) on the same machine, or you could setup a webserver (for html files) and serve this html-file to a webbrowser on a remote machine.

I haven't used rosbride yet, just a C++ websocket library in a system that had nothing to do with ROS.

Hi,

I see two possibilities:

• You could divide your possible velocities in two parts, one part for translation, one part for rotation. For example: set max translation velocity 1/2 of total possible veloctiy, the rotation gets the rest of the possible velocity (or choose another ratio). The sum of both velocities for translation and rotation can't be faster than the real wheel maximal velocity.This lowers your maximal speeds, but you can be sure that every possible velocity command in these limits can be carried out.

• Or you could set both the max. rotation and translation velocities to the physical maximum and observe the resulting wheel velocities after you sum these values. If you max out the possible velocity of a wheel you scale down the velocity of the other wheel so that you keep the ratio of the velocities of both wheels. (Example: your input sets these velocities: left wheel: 100 rpm, right wheel 200 rpm, but your physical max. velocity is 150 rpm, so after scaling: left wheel: 75 rpm, right wheel 150 rpm ) You keep the max. possible velocities if you only translate or rotate, and you keep the shape of the trajectory if a wheel velocity is maxed out. (But the velocity while driving curves can be lower than the velocity that the planner has calculated).

OK, so I misunderstood your problem.

Your function flyToTarget has a never ending while loop, it will not return to the callback and the callback doesn't finish. Get rid of the loop in the callback, you could move the ros loop into your main function. Then publish only a single /cmd_vel per callback.

Have a look at your TF Tree, is there a connection from odom to the robot frames? Maybe you are just missing a static transform from odom to base_plate?

Try changing the rviz 'Global Options' 'Fixed Frame' to my_frame.

In the tutorial you are sending the markers position in the frame my_frame, but rviz is set to display map, and if you have not set any transforms to connect the frame map to my_frame rviz can't calculate the markers position in the map frame.

And you need to add a markers display to rviz (maybe you have already done this, but i can't see it in your screenshot). Just follow the Tutorials Point '5. Viewing the Markers'.

To change the Fixed Frame: Open rviz, klick on map (on the text next to Fixed Frame) and just type the new frame name my_frame. I thought i could choose the frame from a list, but i just built the tutorial program and tried it, but i had to type the name my_frame.

Rviz is often hiding (or it is not recognizable) that you can edit a parameter like Fixed Frame.

I think you are trying to change the Fixed Frame in Global Status, where you get the warning message. Try changing the Fixed Frame under Global Options.

You can write your Arduino software without using rosserial, instead write a minimal serial protocol that is small enough to fit into your arduino memory. Then you need an additional receiver-node on the ROS side, it receives the data over this minimal serial protocol and can construct a big nice standard ROS message out of it.

Where do you get your odometry values from? Does it show any position jumps (in rviz or howeer you can see the values)?

I also played around with the mpu6050 connected to an arduino and wrote myself a small ROS node which connects to an example script from i2cdevlib. You can find this node here:

https://github.com/fsteinhardt/mpu6050_serial_to_imu

Updated: now it reads the orientation, the accelerations and the rotational velocities .