With the ease of getting all the information in a go, people expect their cars to provide a similar experience. They want a solution that can increase their safety and their loved ones. To fulfill that, automakers have already been working on driver assistance systems.
Assistance systems can be divided into two categories:
- Passive ADAS – They can emit auditory and visual signals for alerting the driver about the dangers on the road.
- Active ADAS – They try to avoid accidents by controlling the car without human intervention.
Fig 1. Advanced Driver Assistance System
The driver assistance system depends on a combination of technologies to "assist the drivers in making decisions". IR based camera sensors are the first choice of automakers for monitoring the driver, but it has its limitations. It fails to detect a baby under a blanket or if the baby is outside the camera's FOV (field of vision). Objects behind the front passenger seats can't be entirely monitored as if the child is placed between two rows of a seat or behind the front seats.
Hence, a combination of radar-based occupant detection with a camera-based driver monitoring system can be used as a complete solution for in-cabin sensing.
Driver Monitoring Solution
WHO has estimated that motor vehicle crash kills around 3000 people every day.
The majority of accidents are related to driver error. In an analysis of crashes held in The US between 2005 to 2007, it was found that major reasons for crashes were distraction, sleepiness, and excessive speed.
Infotainment system and smartphones have added to the already demanding activity of keeping the drivers focussed and has maximized the chance of accidents. Although technology has increased the risk of accidents, it can enhance the safety of the driver too.
Automobile manufacturers are continuously seeking innovative solutions to leverage sensor technology to assist the driver. There has been a rapid increase in demand of a driver monitoring system that can sense and monitor the driver's condition and provide timely and useful feedback.
Why is IR camera used?
One of the most helpful technologies in monitoring the driver is artificial vision. It is useful in the development of systems that detect drowsiness and distraction. But these systems have to be enhanced for night conditions. That's why infrared light is used to detect facial expressions in the dark.
How can DMS analyze the driver's state?
Driver's state, such as drowsiness, anger, distraction, can very much influence the driver's behavior. For example, drowsiness can decrease reaction time towards a situation, or anger can lead to risky driving behavior.
Fig.2 Drowsiness detection
While the drivers' exact state can't be measured, it can be predicted from their facial expressions and body postures such as eye gaze, yawn, sitting posture, body movements, etc. Facial and body expressions can be used to detect the driver's vigilant level.
For example, blinking rate, PERCLOS (percentage of eye closure), and yawning can be used to identify drowsiness, whereas eye gaze and head and hand movements can be used in distraction detection. With the advancement in computational power and algorithm design, real-time detection of the driver's condition can be accomplished.
It is worth mentioning that the driver's facial and body expressions can be used to determine the driver's emotions.
Therefore, by monitoring the physiological condition of the driver, their safety can be enhanced.
DMS is usually placed where the driver has direct contact like the steering wheel, the A-pillar, or the rear-view mirror. These contact points allow the system to have appropriate measurements of the physiological variables.
What is the future of DMS?
Soon, there will be a shared responsibility among the semi-autonomous vehicle and the driver. The vehicle would be handling most of the driving tasks but not all.
As semi-autonomous technology becomes more common, there will be more risk of drivers either not understanding the responsibility they currently have or not paying attention to them when they ought to. The driver monitoring system is a necessity until a fully autonomous vehicle hits the market.
That's why for a long time, this area has garnered a lot of interest.
Driver monitoring can be done in two ways:
- Contact Method – This method involves the extraction of physiological features such as breathing, electromyography (EMG), Galvanic skin response (GSR), electroencephalogram (EEG), and electrocardiogram (ECG).
- Contact-less method – This method is done by eye tracking, head movements, facial expressions, or tracking devices. Depending on the device, it could be wired or wireless.
Physiological Sensor signals such as EEG and ECG can offer early detection of fatigue or stress. However, a remote non-contact camera imaging is required for efficient monitoring of the driver.
Fig. 3 Driver monitoring solution
Camera imaging can be operated without having the driver wear any special equipment. Therefore, there is no electrical contact between the driver and the equipment.
It has been established that a camera-based driver monitoring system using physiological signals has some advantages over the sensor-based driver monitoring system because of their non-contact application and not so expensive to use in real-life driving circumstances.
Hence, soon camera-based driver monitoring system would be the preferred solution.
Keeping these points in mind, we have been working on our driver monitoring system, which can provide reliable detection of driver drowsiness and inattention, using a single low-power camera inside the vehicle.
PathPartner Driver Monitoring System uses a hybrid combination of advanced facial analysis algorithms and deep learning models to assess the driver's alertness and focus under challenging environmental conditions. We use a machine learning model for face detection & land-mark regression and shallow & Deep CNN model for estimations and classifications.
Occupant Detection Solution
RADAR (RAdio Detection And Ranging) is being used for decades and can detect the velocity, range, and objects of angles. Radar is lighter than the other sensors and can function in almost all conditions.
They are usually categorized as per their operating distance ranges:
- Short Range Radar (SRR) – 0.2-30m
- Medium Range Radar (MRR) – 30-80m
- Long-Range Radar – 80 - >200 m
Why mmWave Radar?
This radar uses high-frequency electromagnetic waves to map the surrounding. It offers high integration, compact size, low cost, high resolution, and high performance.
It's immune to heat and light. It works at 24 GHz and 77-79GHz. Mm wave antennas are smaller than other alternatives such as ultrasonic, uses less power, and can be easily packed into the vehicle designs.
Their range can be tuned according to the specific requirements. For in-cabin applications, SRR is preferred. The FOV is configured for each car model based on the cabin dimension and position of a radar sensor.
It makes it easier for Radar to detect and distinguish living and non-living things.
How is mm-wave radar used in occupant detection?
NSC has reported that, on average, 39 children under the age of 15 die each year from heatstroke after being left in a vehicle. As a result, the HOT CARS act was created that required the Transportation Department to mandate the all-new vehicle to have a child presence detection system.
Fig 4. Child left in the car
Euro NCAP has also added child presence detection as their roadmap feature. Although the camera-based driver monitoring system can be used to detect the child's presence, it might not function properly if it's not under the camera's view.
Mm wave radar can detect a person irrespective of the lighting conditions. It can also pass through plastic, blanket, drywall, etc. This helps in placing the sensors behind the material inside the car. Mm wave radar sensors can detect vital signs and distinguish one passenger from others.
It's a contactless and non-intrusive technology that is best suited for occupant monitoring.
A perfect In-cabin solution
The next level of monitoring that is coming is in-cabin sensing. Soon the government firms will be stating specifications for in-cabin sensing for auto manufacturers and ratings.
Suppliers are working on software and hardware to meet those needs.
Our driver monitoring system is offered on various platforms so that DMS can be integrated into the instrument cluster using the same processor that drives the cluster. This will help in cutting down the cost.
Also, the small radar can be placed anywhere in the car. The need for accurate in-cabin sensors to monitor the driver and detect the child's presence is a high priority for automotive OEMs and Tier1 manufacturers. The solution has to be highly accurate and in a form factor that can be easily integrated into the vehicle in a non-intrusive way.
Many aspects of the driver's situation can cause accidents such as drowsiness, distraction, depression, driving under the influence, or any other health-related issues. These physiological and mental conditions are very much related to safe driving.
The modern cars are equipped with solutions that can determine the driver's physical and mental state and reduce unexpected crashes. These are low-cost non-contact methods for monitoring the physiological state of the drivers. The camera-based driver monitoring system is both inexpensive and convenient to use. Hence, in the near future, it would be the preferred solution.
Mm wave radar can detect a person irrespective of the lighting conditions. It can also pass through plastic, blanket, drywall, etc. It can also detect vital signs and distinguish one passenger from another. It's a contactless and non-intrusive technology that is best suited for occupant monitoring.
Therefore, a combination of Radar and camera-based DMS can provide a complete in-cabin solution that can take care of the driver's safety and fellow occupants.