NVIDIA Jetson Nano Developer Kit B01

Wi-Fiモジュール（8265NGW）とアンテナ（Econlineshop 3dBi デュアルバンド 802.11a/b/g/n/ac対応 WIFI/Wimax/Bluetoothモジュール用アンテナ MHF4 MHF4-50）を別途購入し使用していますが問題なく動いてます。耐久性は買ったばかりなのでわかりません。

予定通り入荷し、無事に、起動するのを確認しました。 本格的な使用は、これからなので、現時点では星4つという評価です。

J’ai eu le même problème que Thomas plus bas. La carte ne s’allumait pas. Je l’ai retourné et le vendeur devait revenir vers moi, ce qu’il n’a pas fait, le remboursement a pris 2 semaines à ce faire. De plus, il y avait une carte Kubii, un vendeur de carte jetson exactement la même mais pour moins chère. Très déçu du produit je ne recommande pas

Innanzitutto la scatola non è nvidia e questo già mi ha fatto pensare… poi nella descrizione viene specificatamente inserito il codice che si riferisce all’utilizzo della sd, mentre questo modello usa un’emmc da 16 gb, i quali non sono nemmeno sufficienti per installare il sistema operativo ed i pacchetti nvidia… spiacente per il ritardo della recensione ma non ci è voluto poco per capire che il problema non fosse al boot o di tutto il resto ma proprio della scheda arrivata sbagliata

Carrier board logic level converter is defective on all boards; the 3.3V output is 1.8V and randomly switches. Pinmux flashing doesn't fix and if you have any noise in your system the GPIO will flicker on and off rapidly in response to the noise. Inputs require pull down resistors of 1.5kOhm to drain 3mA to ground in order to function and not float high at 1.2V (THESE ARE INPUTS). Sometimes the outputs will randomly change from 1.8V to 3.3V. This is occuring on multiple jetson nano boards and not isolated to one device. The carrier has defective hardware.

Great product

Jetson Nano is great for not only robotics/edge AI, you can use ML for science usage such as medical imaging or environmental data to speed up your workflow. From personal experience even an underclocked dual-core power save mode on the Jetson Nano will still be faster on CUDA AI/ML tasks than a Raspberry Pi 4, however your workflow may vary. If you use AI/ML that isn't optimized for CUDA, in some cases a Pi 4 raw CPU compute can edge out the Nano. I would say if you pair a Pi 4 with any AI/ML accelerator it'll cost more than a Jetson Nano and your mileage is still going to vary. Depending upon how you use a Jetson Nano, for robotics/automation you can actually run four cameras via USB and use the camera interface. Performance wise if you do opt to run a Jetson Nano using USB power, your mileage is going to vary as not all USB power adapters provide a stable voltage which means checking the specs--I reused a Canakit USB power adapter from a retired Pi 3 and never had any voltage warnings but if you plan to run a Jetson Nano hard like a Pi 4 you'll want to use the barrel power adapter for extra power stability when using multiple USB devices+GPIO. Thermal wise I've compared a fanless vs fan equipped Jetson Nano, even under sustained load the heatsink size prevents it from thermal throttling too much. This B01 version has two camera connectors which is geared for stereo imaging however you can run two cameras at a small performance loss and also fixed the networking issue which occurred on the original Jetson Nano A01/A02. From a performance per watt/dollar ratio, if you're going to dive deeper into AI/ML a Jetson NX is more ideal. With a Jetson Nano if you're pushing four cameras and LIDAR it'll require a bit of tweaking to get optimal performance and still remain at about 3.5GB of memory usage.