I recently bought a Zed camera from stereolabs. I am actually interested to the SLAM capability shown on different videos, such as the one in this page: NVIDIA Jetson Partner Story: Stereolabs | NVIDIA Developer
Be fore to buy a Jetson I tried to find some references and if exists a source code or an application capable to do what is shown in the video… Sadly I am not able to find references.
Have anyone tried the SLAM with Jetson and Zed camera?
Hi,
I’m currently working with ZED camera and Jetson TX1.
SLAM is a big field, and there are several SLAM-Projects around.
If you referencing to the software you saw in the video, I didn’t find any source code, provided from
Stereolabs.
In general, with ROS and RTABMAP, I got RTABMAP working.
Stereo labs provides two launch files for ros and a driver. One of the launch files provides tf messages and the other does the depth stuff. To use it with ROS you need to compile the node for the Zed with ROS then create a urdf file for your bot and install slam-gmapping. I followed this site’s instructions substituting the Zed node for the Kinect node. SLAM Robot | #Kahily Projects the map is just fine. And now that I have figured out how to access 2 serial ports on the TX1 my bot should go autonomous as soon as the rain stops. My test area is a bit swampy :) But it all works on the test stand. At least it stops when I stand in front of it now :) Remember with slam-gmapping you have to have encoders or some kind of odometry. If you don’t want to deal with all that look up hector-slam. Its what the drone guys use and doesn’t require odometry obviously since there are no wheels on a drone. A gps and imu will suffice. The Bosch BN0055 imu is well supported check out jetsonhacks.com for a complete tutorial on how to set that up in ROS. A new imu from Microchip is nice no calibration needed. But no ROS support right now although I doubt it would be that hard to convert the BNO055 ROS node to work on the new microchip device. The one I’m using came from Adafruit. With ROS slam just isn’t all that complicated to achieve.