Called PPC-4910, computing comes from NXP’s i.MX8M Plus 1.8GHz quad Cortex-A53 core processor, which has an integrated AI/ML neural processing unit (NPU) operating at up to 2.3Top/s “that focuses on edge AI, vision processing and other multimedia applications”, said Estone. “Paired with the NXP eIQ machine learning software development environment, the system can take the advantage of its integrated hardware accelerator to perform artificial intelligence and machine learning tasks,and supports ONNX Runtime, TensorFlow Lite and DeepViewRT inference engines.”
Intelligent embedded vision applications up to 12Mpixel are supported by two MIPI CSI camera inputs and two image signal processor (ISP) channels.
For real-time control, the computer has a CAN bus interface. “The error-checking mechanisms incorporated with CAN bus also help detect and correct errors during data transmission, which is crucial in applications like medical devices, transportation, industrial HMI, where data integrity is essential,” said Estone.
The PC comes in three versions for different power sources: PPC-4910-MQS for 12Vdc or 24Vac power, -MQP for Power-over-Ethernet (IEEE 802.3at, POE+) and -DQT for 9-24Vdc power – the latter gets a second Gbit Ethernet port too.
Other interfaces include Wi-Fi, Bluetooth, GPIOs, RS-232 or RS-485.
For high ambient lighting conditions, there is a 1,000nit display option.
It measures 258 x 173 x 44.5mm (including the 3mm front glass), mounting is via a 75 x 75mm VESA interface, and operation is over 0 – 60°C.
The board support package covers Yocto Embedded Linux, Android, Qt, Wayland, NXP eIQ Toolkit, “along with inference engines, neural network compilers and optimised libraries”, said Estone, which foresees machine learning applications in image classification, object detection, object tracking, natural language processing and key phrase detection.
The PPC-4910 product page is here, with its software support details here