Review on Enhance Your Development Projects with the NVIDIA Jetson Nano Developer Kit (945-13450-0000-100) by Roberto Garrido

Jetson Nano is not only great for robotics/edge AI, you can also use ML for scientific purposes like medical imaging or environmental data to speed up the work process . From personal experience, even dual-core power saving mode at a reduced frequency on the Jetson Nano still runs CUDA AI/ML tasks faster than on the Raspberry Pi 4, but your workflow may differ. If you're using AI/ML that isn't optimized for CUDA, the Pi 4's raw CPU calculations can crowd out the Nano in some cases. I would say if you pair the Pi 4 with any AI/ML accelerator it will cost more than the Jetson Nano and your mileage will still vary. Depending on how you use the Jetson Nano for robotics/automation, you can actually power four cameras over USB and use the camera interface. If you choose to use a USB powered Jetson Nano your mileage will vary in terms of performance as not all USB power supplies provide stable voltage. There have never been voltage warnings but if you plan to run the Jetson Nano on a hard drive like To use Pi 4 you need to use a cylindrical power supply for extra power stability when using multiple USB + GPIO devices. In terms of temperature, I compared the fanless and fanless Jetson Nano, and the size of the heat sink prevents excessive thermal throttling even under continuous load. This version of the B01 features two camera ports designed for stereo imaging. However, you can use two cameras with little performance loss and it also fixed a network issue that existed with the original Jetson Nano A01/A02. If you want to dive deeper into AI/ML, the Jetson NX is for you. Using four cameras and LIDAR with the Jetson Nano requires a small tweak to get optimal performance and still use around 3.5GB of storage.