I have an Jetson AGX Orin and am curious about some 3d object detection (dope). I found several documentation Websites from NVIDIA which are similar but still different and I find it really hard to find the “right” or newest one.
This link is, I think, obsolet.
So i think i am going with that one…
Also I didn´t get the difference between Isaac Ros dope (here) or the conventional Ros dope (link).
Could you please veryfy or clear my thoughts. Or point me to the right direction… tbh your documentaion is very scatterd. Thanks in advance.
DOPE model inference is available in three different ways depending on which autonomy framework you’re using.
- Isaac SDK has a version built for the Isaac SDK Engine (your first link)
- Isaac ROS has a version built for the ROS2 ecosystem (your second link)
- NVIDIA Research published a version built for the ROS1 ecosystem and provide links to the Isaac ROS version for anyone using ROS2. (your third link)
I would recommend you look at Isaac ROS Pose Estimation (#2) as your best option.
Thanks for the clarification.
We have tried DOPE on Isaac SDK with a real Carter robot (AGX Xavier) and found that the Isaac SDK example works.