NEXCOM partners with Coral from Google
06/11/2020
NEXCOM has announced a series of Coral AI Edge telematics solutions, featuring the Edge tensor processing unit (TPU) accelerator.Coral from Google allows users to quickly, efficiently and securely perform inference at the edge, avoiding the additional time and cost involved in uploading to a centralised cloud. The NEXCOM and Google collaboration creates all-in-one artificial intelligence (AI) ecosystems that support several specific use cases in the transportation and public works sectors, including object detection and condition monitoring.
The on-board Google Edge AI accelerator is an application-specific integrated circuit (ASIC), small but sophisticated enough to perform inference with big data and deep neural networks. Its compact mPCIe form factor allows it to be energy efficient, space saving and of lower cost, especially compared to competitors’ solutions, it is claimed. Developers and system integrators that are familiar with other Google Cloud services will enjoy seamless end-to-end infrastructure integration.
Additionally, Google Coral TPUs use proprietary TensorFlow Lite, an open-source machine learning inference framework, to train new or pre-existing models for deployment on any of NEXCOM’s Google Coral AI Edge telematics solutions. Using the framework locally on its platforms means lower latency, time savings and greater privacy, enabling users to complete modelling and learning tasks more quickly and securely.
NEXCOM’s array of Google Coral AI accelerator solutions emphasise deep learning for increased transportation safety and efficiency. For instance, the telematics gateways’ PoE ports support IP cameras for continuous on-board surveillance, yet also combine with optical character recognition (OCR) for automatic number plate recognition (ANPR) capabilities. In the same vein, people counting and tracking functions integrate with GPS to provide valuable data such as load factors and passenger usage, without incurring unnecessary human labour expenditures. Smart sensors provide another layer of safety by learning to detect foreign objects and thus avoid accidents. The control centre can utilise this information to manage traffic at intersections. All of the aforementioned examples are just a few of the many ways to partner AI with transportation to create a more advanced and more unified smart city.
