Multiple Kinects?
Is it possible to use more than one Kinect to gather multiple views around an object at the same time?
ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange |
Is it possible to use more than one Kinect to gather multiple views around an object at the same time?
It is absolutely possible. I have run 4 ASUS sensors (Kinect) from one computer. Each Kinect needs to be on its own bus, or you will have issues getting all the data through. You need to give each Kinect a unique camera name in the launch file and create separate frames.launch files for each one. I don't have access to the launch files right now to give you an example, but I can get you an example launch file tomorrow.
EDIT
Here is an example launch file that I used recently to launch two Kinects. This should be a decent template to get you going. This launch file is specific to my system (the "device_id" values are specific to my bus configurations), so it probably won't work by simply copying and pasting. Also, I was running this on Ubuntu 11.10 with ROS Electric:
k1_2.launch
<launch>
<arg name="debug" default="false"/>
<arg if="$(arg debug)" name="launch_prefix" value="xterm -rv -e gdb -ex run - args"/>
<arg unless="$(arg debug)" name="launch_prefix" value=""/>
<node pkg="openni_camera" type="openni_node" name="openni_node1" ns="camera1" launch-prefix="$(arg launch_prefix)" output="screen">
<param name="device_id" value="004@02"/>
<param name="rgb_frame_id" value="kinect1_rgb_frame" />
<param name="depth_frame_id" value="kinect1_depth_frame" />
<param name="use_indices" value="false" />
<param name="depth_registration" value="true" />
<param name="image_mode" value="2" />
<param name="depth_mode" value="2" />
<param name="debayering" value="2" />
<param name="depth_time_offset" value="0" />
<param name="image_time_offset" value="0" />
</node>
<node pkg="openni_camera" type="openni_node" name="openni_node2" ns="camera2" launch-prefix="$(arg launch_prefix)" output="screen">
<param name="device_id" value="003@05"/>
<param name="rgb_frame_id" value="kinect2_rgb_frame" />
<param name="depth_frame_id" value="kinect2_depth_frame" />
<param name="use_indices" value="false" />
<param name="depth_registration" value="true" />
<param name="image_mode" value="2" />
<param name="depth_mode" value="2" />
<param name="debayering" value="2" />
<param name="depth_time_offset" value="0" />
<param name="image_time_offset" value="0" />
</node>
</launch>
Check out the answer to this question: http://answers.ros.org/question/28826/multiple-kinects-overlapping-fov-single-computer/
even better script here http://answers.ros.org/question/96071...
Yes, but there may be bandwidth problems if you connect them both to the same USB port.
Asked: 2013-03-12 11:54:44 -0500
Seen: 2,946 times
Last updated: Mar 13 '13
problem building my map using SLAM
Kinect calibration "Unknown OpenCV format yuv422"
robot_self_filter + rgb pointcloud
Turtlebot follower demo not following
Slow Kinect frame rate in openni parameters
could not find executable file
kinect green light keeps freshing. Seems no driver
Can monitored planning scene subscribe to multiple point clouds