AEKD-AICAR1 Automotive AI on the edge for car state classification

The AEKD-AICAR1 is a versatile system based on a long-short term memory (LSTM) recurrent neural network (RNN), which can provide a car state classification: car parked, car driving on normal road conditions, car driving on a bumpy road, car skidding or swerving.

The innovative idea in the AEKD-AICAR1 is to define an ECU detection node with an embedded artificial intelligence processing.

The AEKD-AICAR1 houses an SPC58EC chorus 4Mbytes microcontroller, which can acquire discrete acceleration variations on a three-axis reference system.

The AEKD-AICAR1 system represents a reference for the automotive AI on the edge processing.

It is possible to replace easily the sensor with another one that belongs to the ST MEMS family. It is also possible to modify the neural network and/or retrain the neural network. The new neural network is converted into a library executable by the MCU using SPC5-STUDIO-AI.