Machine learning Artificial intelligence applications running on IoT devices are opening up a new wave of highly valuable IoT opportunities that generate stronger operational insights, offer faster processing and decision making from real time data, and elevate the IoT market to tackle industrial operational, site safety, and personal health analytics solutions.
We are entering an innovative technology inflection point where the intelligence of IoT devices is increasing dramatically due to the inclusion of artificial intelligence and machine learning solutions running right inside the endpoint devices. Sitting between raw control & sample IoT devices like temperature sensors, bulbs, and locks, and advanced compute systems like the autonomous systems running in cars and agricultural equipment is Edge compute. By deploying ML and AI solutions on IoT Edge hardware construction site safety can be improved by using stand-alone hard-hat detection, industrial machinery can be monitored for noise and vibration patterns to help not only detect early onset of failure but also make suggestions as to the root cause and repair.
Edge compute is an opportunity to deploy advanced decision-making opportunities to system design architecture by adding on-site logic that can even up-skill existing basic control & sample IoT devices. In many cases low power endpoint devices do not possess enough compute capability to run advanced AI and ML algorithms on the input data that they are capturing. System architectures can be updated by relaying the raw data to an edge compute device performing as a gateway offering a strong solution by running advanced AI and ML applications running locally at the deployment site within the gateway due to its advanced compute capabilities. Simple raw data forwarding from the endpoint to the gateway provides an architectural separation between the compute capability requirements of an endpoint and its upstream gateway while at the same time bringing a more robust system design to the solution by adding consistency through outage tolerant processing.
As with all IoT equipment the secure management and interaction of Edge compute equipment is critical to keeping your systems online and ensuring that the critical data that feeds your wider systems and on which you make operational decisions is consistently trusted. Pelion is all about scalable enterprise solutions to manage and interact with your endpoints and connected gateways in the world of IoT. Simple, safe provisioning of devices using mutual authentication, firmware remote update management of the applications running on devices, and secure IoT data messaging upstream that are the hallmarks of an enterprise grade IoT solution are the primary virtues of Pelion. With AI and ML techniques playing more and more of a role in the function of endpoints and gateways, Pelion has teamed up with Amazon AWS to show how AWS’s SageMaker ML development environment and Pelion-controlled gateways can be used to manage, update, and securely apply AI & ML solutions to your system requirements
AWS SageMaker is a data scientist’s dream when designing ML solutions. It provides a uniform, elegant IDE for the data scientist to create, develop, and train specific ML models for specific tasks. Furthermore, the SageMaker NEO tool from AWS intelligently compiles the developed model into highly-optimized native code for specific platforms – one such example is Nvidia’s Jetson Xavier NX development board.
To learn more about how Pelion’s secure IoT management platform deploys ML models to edge devices and integrates with their results take a look at our explainer video, or to see the solution running on the Nvidia Jetson Xavier NX development board watch the full demo walkthrough.