Precision PCB Fabrication, High-Frequency PCB, High-Speed PCB, Standard PCB, Multilayer PCB and PCB Assembly.
The most reliable PCB & PCBA custom service factory.
New sensor technologies can promote innovation in driver assistance systems, car automation, car networking, and mobility services. With the improvement of the level of driving automation, the complete revolution of the in-vehicle system has made the auxiliary system outside the vehicle more powerful, thus creating an overall driving experience. This article mainly describes how short-range radar sensors empower some automotive cockpit sensing applications, namely driver monitoring systems and occupant monitoring systems.
Human-computer interaction (HMI) is becoming an area where automakers are pursuing differentiation. Automotive human-computer interaction technology first emerged in 2015, when it only used infrared cameras and MEMS tactile feedback systems to achieve simple gesture sensing. Today, this PCB technology has been developing towards a fully personalized and super-large digital display. Byton's M-Byte 48-inch co-pilot display and Daimler's MBUX system are typical examples. These automotive instruments will completely change human-vehicle interaction.
The advancement of sensors in miniaturization, in-dashboard processing, energy efficiency, and ease of integration will promote the development of newer and more advanced technologies such as radar sensors and time-of-flight sensors. In addition, sensor fusion heralds the future direction of development, such as the combination of sound and gestures to reliably predict the user's target action, illuminate the display button when the user approaches, and distinguish the input information of the driver and the occupant. The required information, aesthetic design, environmental factors, and computational cost will define the technology for a specific use case. There are many related use cases, including but not limited to comfort applications such as gesture sensing and passive safety applications.
According to statistics from the World Health Organization, approximately 1.3 million people die in traffic accidents every year1, and 73% of these accidents are caused by human error. According to the National Highway Traffic Safety Administration of the United States, more than 50 children die each year because of heatstroke stranded in the car. 2 The EU and ASEAN’s new car acceptance plans are already taking steps to introduce child presence detection systems and driver monitoring systems. The U.S. Automobile Manufacturers Union has signed a voluntary agreement on the rear seat reminder system in September 2019; 3 At the same time, the United Nations Economic Commission for Europe Regulation No. 16 comprehensively describes the seat belt reminders and restrictions in countries such as the European Union and Japan. System function standard. 4 Therefore, driven by laws and regulations, innovative passive safety applications in the cockpit are bringing changes to road safety.
Radar processing-a new transformation
The working principle of radio detection and ranging (radar) is to emit electromagnetic waves and then receive the electromagnetic waves reflected by the object. Most of the information related to the object is hidden in the phase and frequency of the electromagnetic waves received by the radar. This information can be easily extracted and used to locate the target's basic parameters such as distance, angle, and speed. Through the conversion of two-dimensional and three-dimensional signals (such as distance Doppler or micro-Doppler), more information can be obtained to understand subtle body movements, and even chest movements caused by heartbeat and breathing. For classification, radar point cloud images can also be used.
Some of the unique advantages of radar are that it can perceive objects from a shape angle without relying on light conditions, can maintain data privacy through internal coded information, and can work in line of sight and non-line of sight conditions. But its application depends on the specific use case. Some examples will be discussed below.
Driver monitoring system
For driver monitoring systems, the most advanced sensor technology currently is a 2D camera. These cameras are generally installed directly in front of the driver on the steering wheel or dashboard near the speedometer and tachometer. In situations where it is extremely necessary to understand the physiological health of the driver as a whole-for example, in a traffic jam scenario, it may be necessary to adopt a multi-sensor combination method to achieve level-2 or higher autonomous driving. Table 1 summarizes some methods applicable to different use cases.
In the standard radar vital signs signal processing process, radar interferometry technology is needed to monitor the changes in the phase of the detected target over time. 6, 7 After the fast Fourier transform (FFT) processing of the distance, the conventional 1D CFAR technology can be combined with the peak search on the range spectrum, or the peak-to-average power ratio (PAPR) The ratio of the peak to average power in the slow time domain in the range bin of the potential target is used as an indicator to select the potential target. For a stationary target, the FFT peak value is close to the average value of the FFT spectrum in the slow time domain; and if it is a vibrating target, such as heartbeat or breathing, the average value is very small, making the PAPR large.
After pre-selecting the target distance interval, the vital signs Doppler detection can be performed in two ways: 1. Estimate the standard deviation of the IQ data in the slow time domain to see if it is within the range of the specified value; 2. If it is within the frequency of the vital signs If there is no energy peak in the range (0.2-3.3 Hz), the distance spectrum is used for measurement. Because white noise can make the wrong signal a valid signal, Doppler detection is a very important step before passing the signal through a band-pass filter to filter out static targets.
After completing the vital signs detection, the ellipse reconstruction algorithm is used to correct the IQ data that reaches the above-mentioned standard distance interval to eliminate the offset, phase and amplitude imbalance caused by hardware defects. By mapping the ellipse onto a perfect circle, ellipse reconstruction can help eliminate these amplitude and phase shifts. 8 Figure 2 shows the IQ signal reconstructed when the ellipse reconstruction algorithm is used for normal vital signs targets and random body movements interfere with the reconstruction.
Next, the obtained signal phase is used to reconstruct the original true phase of the wave from its 2π multiples through the phase unwrapping module. For phase jumps greater than -π or +π, 2π needs to be added or subtracted, respectively. The unfolded phase contains the displacement signal:
Among them: λ is the wavelength of the carrier, and ϕ(t) is the phase extracted in the slow time domain.
The resulting displacement signal contains the superposition of the respiratory signal and the heart rate signal. Let the displacement signal pass through a band-pass filter to estimate the respiratory frequency when the start frequency and stop frequency are 0.2 Hz and 0.4 Hz, respectively; and the heart rate is estimated at 0.8 Hz and 3 Hz, respectively. 10 There are many ways to estimate respiratory rate or heart rate, including:
1. The distance spectrum estimation technology requires fast Fourier transform (FFT) on the filtered displacement signal. Through the peaks of the heart rate and respiratory frequency in the FFT distance spectrum, the heart rate and respiratory frequency can be estimated separately. Figure 3 shows the frequency estimation of vital signs using the distance spectrum analysis method.
2. Estimate the respiratory frequency and heart rate by statistically filtering the peak value in the time-domain displacement signal. Figure 4 estimates the frequency of vital signs by performing peak statistics on filtered time-domain data. The red triangle represents the peak value detected in the heartbeat signal window.
Cockpit sensing is an emerging market, and it is expected to make strides with the introduction of local regulations. Radar is considered to be a highly potential technology that can be used to solve many problems including passive safety applications, such as stranded child detection and presence sensing. Innovative signal processing and deep learning technologies will bring the reliability of these applications to a higher level, so as to achieve a perfect balance between computational cost, the degree of information required for specific use cases, and system power consumption. In the future, the PCB multi-sensor fusion method should be able to create a more complete and reliable system by realizing sensor redundancy.
