Capsense Based Driver Grip Detection for Drowsiness Analysis

Safety is the utmost important factor, pertaining to automobile driving. There are several methods already available for estimation of driver drowsiness level detection in cars for safety. Many of the high-end cars uses, car lane keeping information, driver's eye analysis, driving pattern etc for estimating driver's fatigue. With the advent of deep neural network algorithms, the estimation could be incorporated easily on the main vehicle computer present capable of deep learning from the sensor data. Grip detection and steering pattern serve as crucial datasets for accurate detection of this problem.

This example project is very simplistic form of the actual holistic solution. Due to limitation of the hardware and unavailability of training dataset already available associated with drowsiness level and grip detection, already; I am using a very basic approach.

The PSoC Creator project attached is created is designed by modifying the Proximity Code Example(CE214023). Apart from the Code example we have extra Timer module for generating ticks for issuing alarm level. Also a UART component to send serial commands, which indicates alerts.

Steps For Testing

1)Open the Project with PSoC Creator 4.4

2)Build and program the code to CY8CKIT-041-41xx

3)Connect the USB cable of the Kit a PC

4)Open any Terminal Software (HyperTerminal, TeraTerm etc). Connect to the serial port associated with the kit. Make sure the baud rate is 152000

5)If the hands are off, with respect to the kit, you can see alert messages displayed on the Serial Terminal.

If there is no hand on the steering wheel, there is an alarm. The code flow of the project is very simple. If the raw value of the Capsense is greater than a threshold, then the System assume that grip is proper. Now the Capsense’s touch threshold has been set such that when the grip is loose , the Capsense would be inactive. In this case there are alerts generated. The alert levels alert, medium alert, high alert etc are displayed on the UART terminal.

Future Advancements

As stated earlier, this is a very small part of the big solution. The real idea is to track heart rate, driver’s eye movement pattern, steering wheel grip , vehicle lane deviation etc to a machine learning model and predict the drowsiness level of the driver. An appropriate action is taken (such as alarm switched on), based on the result of the analysis. In the actual application the Capsense-touch panel should be in-cooperated with steering wheel. Also, the hand gestures while driving must be also captured and sent to the Machine Learning model. If the PSoC family already have BLE or WiFi enabled the grip-related data can be easily shared with the central gateway or the high-performance Vehicle Computer.