On Dec 5th at 8am PDT, we will be hosting a free webinar on how to use your own CUDA code with NITROS and Isaac ROS.
CUDA provides a programming model to develop accelerated computing functions on a GPU. NITROS (NVIDIA Isaac Transport for ROS) is a technology to enable streaming through accelerated computing ROS 2 graphs. Combining the two allows for your own CUDA nodes to take advantage of NITROS with other Isaac ROS nodes and your own CUDA nodes. For more details refer to the CUDA with NITROS docs.
A practical example of this is using your own DNN, with the DNN inference node. Your own DNN may require custom pre-processing of the image into tensors for inference, and custom post-processing to interpret the DNN’s results into a format used by your application. All of this pre and post processing can entirely be done in CUDA to improve the performance and reduce latency to receive the DNN predictions
Register for this webinar to learn how to use your own CUDA code in combination with NITROS.