The system was distributed to multiple computational units Jetson Nano due to the high requirements for processing data from each camera. Python wheel of MediaPipe for Jetson Nano compiled with CUDA support can be found here: Google's MediaPipe (v0.8.9) and Python Wheel installer for Jetson Nano (JetPack 4.6) compiled for CUDA 10.2
@Article{s23094219,
AUTHOR = {Vysocký, Aleš and Poštulka, Tomáš and Chlebek, Jakub and Kot, Tomáš and Maslowski, Jan and Grushko, Stefan},
TITLE = {Hand Gesture Interface for Robot Path Definition in Collaborative Applications: Implementation and Comparative Study},
JOURNAL = {Sensors},
VOLUME = {23},
YEAR = {2023},
NUMBER = {9},
ARTICLE-NUMBER = {4219},
URL = {https://www.mdpi.com/1424-8220/23/9/4219},
ISSN = {1424-8220},
DOI = {10.3390/s23094219}
}
All camera nodes and the master must be set to distibuted ROS graph mode.
Launching the dirstibuted system with three camera nodes:
roslaunch jetson_camera_node start_distributed_graph.launch
Start the aggregator after the camera nodes started:
rosrun jetson_camera_node main_node_processor.py