Our customer needed a device that was powerful enough for their machine learning application. The AI was to be trained directly on the edge device. By applying the AI (Artificial Intelligence) in the field level, the running costs for data transmission are reduced and at the same time a complete data pool of the individual plant is made available to the AI.
The plant status data required by the AI, such as acceleration, temperature, brightness or air pressure, are recorded by means of sensors. The SYS TEC electronic development team integrated these directly into the device and created the option of connecting further sensors via CAN bus, Bluetooth or Sub-GHz. Ethernet, WiFi and LTE interfaces are available for communication.
The required high computing power within the edge device is provided by the NVIDIA® Jetson Nano™ module, which is designed for use in embedded IoT applications. It is plugged as a System on Module (SOM) onto the carrier board developed by SYS TEC electronic.
The training of an artificial intelligence, e.g. with neural networks based on tensor flow, can thus be carried out directly in the field. The resulting intelligent gateway is capable of analyzing the sensor data directly in the field and forwarding it in a preprocessed form. It meets the high customer requirements. The data connection of the device to a central cloud instance can be omitted.