Ten years ago, NVIDIA CEO Jensen Huang started to transform the GPU from an extraneous graphics booster for a niche market of PC gamers into general purpose graphics chips that could be used for anything, from rendering to simulation to CAD workflows.
But he also saw the importance of GPUs to deep learning tasks in artificial intelligence (AI), and NVIDIA’s explosive growth in value has left AMD, Intel, ARM and others scrambling to catch up. Parallel processing on hundreds or thousands of cores embedded in the company’s programmable GPUs turned out to be ideal for offloading numerically intensive computations.
GPUs are ideal at processing floating-point arithmetic at much higher rates than conventional CPUs. They also turned out to be ideal for processing algorithms that could be structured as streaming computations, especially streaming computations that are not only both highly parallel and numerically intensive, but do not reuse input data.
NVIDIA and Komatsu Smartconstruction
At GTC Japan, Huang made the announcement of the company’s partnership with Komatsu’s Smartconstruction job sites. Komatsu rolled out the Smartconstruction job site in 2015 as a response to Japan’s shrinking skilled workforce, helping job site workers connect with information from construction machines and terrain. Since its launch, Komatsu has deployed Smartconstruction at more than 4,000 job sites in Japan. Smartconstruction centers around a new type of dealer employee called a Technical Service Expert, known as a TSE.
The TSE is a gatekeeper to all the Smartconstruction machines. With a simple phone call to a Komatsu TSE, you can schedule a demo of a GPS/GNSS machine control system. If you choose to purchase it, the TSE will set up your first job site, train your operators, guide you with best practices, digitize your site plans in machine monitors and perform a UAV survey of your site.
The GPUs will communicate with drones from Komatsu partner Skycatch, and collect 3D images, visualize site conditions and create accurate terrain data. This data will in turn be processed and used by construction machines and jobsite workers for visualization in an application to be created by IoT management software company and Komatsu partner, OPTiM Corp.
Indicating a future packed with autonomous large-scale construction machines, Huang noted that, “Future machines will perceive their surroundings and be continuously alert, helping operators work more efficiently and safely. The construction and mining industries will benefit greatly from these advances.”
What is NVIDIA Jetson?
NVIDIA Jetson is an AI platform that acts like the mind of Komatsu’s heavy machinery for jobsites. Jetson works in tandem with NVIDIA’s cloud technology and is installed on construction machines. Once installed, it can provide 360-degree images, and recognize other machines and workers on and nearby a given jobsite. This allows it to work as a ground-traffic controller, coordinating machines and people on site to prevent collisions and other accidents.
Jetson is physically tiny, about the size of a credit card, and delivers AI computing at the edge.
The Jetson TX2 uses stereo cameras installed in the construction equipment to recognize job site conditions as they change in real-time to give instructions to machine operators, who are human.
In the future, Komatsu, NVIDIA and its partners plan on not only ceding automatic control of devices to Jetson, but also rendering and creating simulations of construction and mining jobsites.
Are we racing too fast towards integrating AI and automation into all areas of global industry? What do you think?