Solution Description
With today's focus on increasing vehicle safety, new technologies are making it easier to protect drivers and their passengers. This Advanced Driver Assistance System (ADAS) provides a solution that detects vehicle movement and road conditions and warns the driver of uneven and bumpy road surfaces or if the vehicle is skidding or sliding.
Built around a general-purpose SPC58EC automotive microcontroller with 4 Mbytes of Flash memory, the system uses an ultra-low-power AIS2DW12 3-axis motion sensor to acquire vehicle motion variations as well as information on current road surface conditions. The acquired data is passed to an LSTM (long short-term memory) recurrent neural network running on the MCU that classifies the driving conditions and generates a precise vehicle state prediction model that is displayed on an LCD touchscreen.
To start your development first download the AutoDevKit and then, to obtain a pre-trained neural network, developers can use one of the many available frameworks (Tensorflow, Lasagne, etc.) with an external IDE (Google Colab) to create, train, and validate (using truth tables) an LSTM recurrent neural network. Once the LSTM RNN is created, it must be exported into an H5 file format so that the Artificial Intelligence (AI) plugin (SPC5-STUDIO-AI) running on the SPC5 Studio development environment (SPC5-STUDIO) can convert it into C code in order to make it executable and load it in the SPC58 MCU.
-
Key Product Benefits
AIS2DW12 - Ultra-low-power 3-axis digital motion sensor
-Ultra-low power consumption
-Embedded self-test feature
-Small thin plastic LGA package
-Two supply voltages (1.8V or 3.3V)
SPC58EC80E5 - 32-bit MCU for automotive applications
-4 Mbytes of Flash memory
-Supports the ASIL-B safety standard in compliance with ISO 26262
-Rich set of communication interfaces
-Embedded Hardware Security Module (HSM) with dedicated Flash memory
-
All Features
- Vehicle movements are analyzed over 6-second time periods
- Dedicated Long-Short Term Memory (LSTM) recurrent neural network for time series analysis
- AEC-Q100 qualified ICs
Key Benefits
- Artificial Intelligence on the edge for automotive applications
- Run a pre-trained neural network on a ‘simple’ MCU
- Sensor accelerations analyzed in real time
- Dedicated Long-Short Term Memory (LSTM) Recurrent Neural Network for time series analysis
- Four car states are recognized:
- Car parked or stopped
- Car driving on normal conditions road
- Car driving on a bumpy road
- Car skidding or swerving
