Nvidia GPU Enables Drone Inspections to Utilize AI

droneAutomated inspections by an intelligent drone with deep-learning capabilities was recently demonstrated at the European GPU Technology Conference. Aerialtronics, a Dutch UAS manufacturer; Neurala, a creator of deep-learning software; and NVIDIA, the world leader in GPU accelerated computing jointly developed the prototype.  AI will take drone autonomy to the next level in the near future where “drone inspections can be run at the push of a button” said Roger Matus, Vice President at Neurala.

Intelligent drone use cases include inspections of cell towers, bridges, buildings, and wind turbines, and other types of physical infrastructure. Using Nvidia Jetson platform with high-performance low-energy computing for deep learning analytics along with computer vision, intelligent drones can access hazardous locations to handle complex inspection tasks with more accuracy and efficiency. Finding and recognizing objects in real time during flight advances drone inspections to the next level.

With deep-learning autopilots enhancing the data capture, the amount of data collected by intelligent flying platforms will continue to increase at levels previously not considered. Now add the next technology approaching mainstream, autonomous vehicles.  With all this collected data incoming, the requirement to analyze and monetize pixel based data becomes paramount. Nvidia has proven for ten years that Graphics Processing Units (GPU) and Accelerators are required to off-load algorithmic and computational work from the CPU in non-transactional application environments.

DroneData servers support the entire product line of Nvidia Quadro Graphics Processing Units for visual computing applications and Nvidia Tesla Accelerators for compute only applications such as deep-learning and AI.  With no upfront cost and no rental term commitments, DroneData GPU dedicated and accelerated servers handle all compute intensive applications from the smallest to a whopping 20,000 GPU computing cores in a single server.