I don't get it. The tracker was used separately (additionally) to provide IMU data to the visual odometry. But this presentation is of 28th May 2019. Why didn't he use D435i that has a built-in IMU ? Shouldn't that provide both IMU and Visual Odometry together as a single unit?
Adding the T265 provides VIO based odometry topic to the system making tracking much more accurate than just IMU that is prone to drift as shown on the first video
the tracker outputs the visual odometry using it's onboard asic to compute whereas the d435i outputs the imu data that would then need to but fed into an algorithm running on the host device CPU or GPU
Can I use these products for commercial applications? Can I make a robot with an Intel T265 and Jetson Nano and just sell it? Or will that require licenses and stuff?
@@TheRockeyAllen you need to know the regulations for the field where you are deploying your robotics solution. You need to look for ISO standards and such regulatory documents. If you are selling a surgical robot to a hospital then you need to follow medical standards. If you are selling a drone to consumers then you need to follow aviation regulations, etc.
A T265 overview is here, basically a paper version of this talk, with links to the T265 ROS node source code: dev.intelrealsense.com/docs/intel-realsensetm-visual-slam-and-the-t265-tracking-camera SLAM with D435i setup guide for ROS: github.com/IntelRealSense/realsense-ros/wiki/SLAM-with-D435i RTAB-Map (underlying SLAM library) source code: github.com/introlab/rtabmap This also needs the Point Cloud Library: github.com/PointCloudLibrary/pcl
Hi I have some questens. How did you make the Sensorsystem? Which hardware did you use, can I bild it with a Jetson Nano and a Akkupack? Is there a German Person to talk?
I tried this with my Realsense d435i on a jetson nano. It lost track easily, i even move it slowly with my hand. Anyone have suggestion to fix the problem?
@@Frankx520it got better after i set the parameter recommended here wiki.ros.org/rtabmap_ros/Tutorials/HandHeldMapping Dont use the launch file included on realsense sample folder, read RTabMap wiki and github so you can tune it better. Goodluck~
Closest would be the HTC Vive Ultimate -> Calibrate -> Run it in STEAM VR in Headless Mode -> Use OpenXR package in Python, installed via pip, to read the pose
![Megan Thee Stallion - Neva Play (feat. RM) [Official Video]](http://i.ytimg.com/vi/TpYTyAaTRts/mqdefault.jpg)
08:06, available SLAM solutions
25:33, global shutter for rapid moving platforms
The issue is the extremity limited range. If it worked well.at 30 meters then it would be a lot more useful. Also has issues in sunlight reflections
I don't get it. The tracker was used separately (additionally) to provide IMU data to the visual odometry. But this presentation is of 28th May 2019. Why didn't he use D435i that has a built-in IMU ? Shouldn't that provide both IMU and Visual Odometry together as a single unit?
Adding the T265 provides VIO based odometry topic to the system making tracking much more accurate than just IMU that is prone to drift as shown on the first video
the tracker outputs the visual odometry using it's onboard asic to compute whereas the d435i outputs the imu data that would then need to but fed into an algorithm running on the host device CPU or GPU
Nice video. At first I thought it was "Realsense with Jake Gyllenhaal"
SLAM = Simultaneous Localization And Mapping
Where is a link to the next speaker
How does this software account for moving objects, such as other people walking around in that office?
Does it work on D435 and L515 cameras?
Can I run a drone with jetson nano and the stack of t264 and d400 cameras and use them for indoor navigation ?
Can I ask what's kind of SLAM this project used?
Can I use these products for commercial applications? Can I make a robot with an Intel T265 and Jetson Nano and just sell it? Or will that require licenses and stuff?
depends on who you sell it to :)
@@robnierreyes7197 oh... where can I find more information on this?
@@TheRockeyAllen you need to know the regulations for the field where you are deploying your robotics solution. You need to look for ISO standards and such regulatory documents. If you are selling a surgical robot to a hospital then you need to follow medical standards. If you are selling a drone to consumers then you need to follow aviation regulations, etc.
Basically buy a used Xbox kinect and gain color and depth.. instead of just depth.. you can buy a kinect for 3 dollars
That's great, Where can I find the source code or sample?
I second the motion.
I would like to have complete documents and source code files to read. Where I can get them ?
A T265 overview is here, basically a paper version of this talk, with links to the T265 ROS node source code: dev.intelrealsense.com/docs/intel-realsensetm-visual-slam-and-the-t265-tracking-camera SLAM with D435i setup guide for ROS: github.com/IntelRealSense/realsense-ros/wiki/SLAM-with-D435i RTAB-Map (underlying SLAM library) source code: github.com/introlab/rtabmap This also needs the Point Cloud Library: github.com/PointCloudLibrary/pcl
@@olawlor Thanks a lot!
Good presentation
Hi I have some questens. How did you make the Sensorsystem? Which hardware did you use, can I bild it with a Jetson Nano and a Akkupack? Is there a German Person to talk?
Hey, ich habe ein ähnliches Vorhaben wie du. Hast du dein Projekt inzwischen beenden können?
I tried this with my Realsense d435i on a jetson nano. It lost track easily, i even move it slowly with my hand. Anyone have suggestion to fix the problem?
I just bought a D415 for my Jetson nano robot, and I saw your comment... So, is your tracking get any better now?
@@Frankx520it got better after i set the parameter recommended here wiki.ros.org/rtabmap_ros/Tutorials/HandHeldMapping
Dont use the launch file included on realsense sample folder, read RTabMap wiki and github so you can tune it better.
Goodluck~
@@_aawawaa_ COOL! I am also trying to add two encoders, and IMU for the robot car. I hope I can get a good result : ) Tx for the link!!!
Also the current cameras has no build for pi4
Shukran habibi
The sensor was not correctly design. Use space sense without camera
Steve Carell explaining SLAM...
3 years later... are there any better solution on the market??? I can't believe Intel or someone else didn't come with a single camera doing everthing
Closest would be the HTC Vive Ultimate -> Calibrate -> Run it in STEAM VR in Headless Mode -> Use OpenXR package in Python, installed via pip, to read the pose
👏👏👍
👌👍❤