NVIDIA Jetson TX1 is a promising piece of hardware. It is designed to accelerate the development of AI cognition. This not only includes self-driving vehicles, but drones and robotic systems as well.
In much the same way GPGPU complimented the CPU, humans will gain a tremendous boost in productivity from machine learning AI systems.
For the GPU developer, The Jetson is a supercomputer module the size of a credit card. It consists of a 256 core NVIDIA CUDA processor built in the NVIDIA Maxwell architecture. Jetson combines 64-bit CPUs, deep learning algorithms, computer vision processing and GPU computation in an embedded form factor.
The Jetson comes in three version: The Jetson TX1 module and two developer kit. All are very low cost. $192 for the TK1 dev kit, $599 for the TX1 dev kit, and $435 for the compute module. The Dev kits are delivered with a complete suite of development and profiling tools with support for cameras and other peripherals right out-of-the-box. Meaning you can begin incubating your embedded AI projects without breaking the bank. Nvidia’s developer portal is a great place to kickstart your project.
I expect to see Nvidia embedded AI solutions powering a new generation of devices. From areal drones, to home security appliances to advanced multi-factor authentication and intrusion protection systems. The convergence of low power high performance AI will bring great opportunities.
However, lucking in that promise is the potential for easy API exploits and privacy breaches. For example, an army of IoT devices were uses to launch DDoS attacks against Krebs. Think of the impact if thousands of supercomputers on a chip were used instead. With more sensors embedded in our homes and personal devices, data mining private information will become a big issue. Look at this graphic from a mapD demo. It shows
individual’s names, their donations, and if you zoom in enough, their location. All easily viewable online.
It will be interesting to see how we navigate these two issues in the future.