By Jim Toal, Senior Manager, Product Marketing at Vishay
Whether driving at high speed on Germany’s Autobahns or on California’s Highway 1 as it twists and turns along the Big Sur coast, drivers cannot afford to take their hands off the wheel or take their eyes off the road. A distracted driver is an unsafe driver, which is why car manufacturers are designing in features to keep the driver’s eyes on the road and not on the large center console display.
To achieve this, they are developing new Human-Machine Interface (HMI) technologies which include heads-up displays, speech recognition and gesture control, as shown in Figure 1. Functions currently using simple swiping gesture controls include changing the radio channel, answering a phone call, playing the next song on a playlist, menu scrolling, navigation screen zooming, opening and closing the sunroof, and overhead dome light control. Gesture control and the sensors that enable the feature are available now mostly in premium vehicles.
A market study conducted in 2018 by Global Market Insights forecasts the automotive gesture recognition market to grow in value from $1.1m in 2017 to $13.6bn by 2024. The market for hand gesture recognition is projected to be greater than $8.6bn by 2024. Compounded annual growth between 2018 and 2024 for both North America and Europe is projected to be greater than 40%. The report states that safety regulations introduced by various governments are improving product penetration. But tough challenges, including high cost and complexity, and the difficulty of integrating new HMI technologies into system designs, will nevertheless raise barriers to the implementation of gesture recognition in cars in the years up to 2024.
Fig. 1: Gesture control enhances the safety of sophisticated user interfaces in the car
Components for Optical Gesture Systems
Optical sensors such as Vishay’s VCNL4035 may be used to detect swiping hand motions. An optical gesture system can comprise an integrated proximity sensor that can pulse up to three emitters in quick succession, combined with external infrared emitters to create a zone in which a gesture can be performed.
In fact, for simple left- and right-swipe and zoom-in and zoom-out motions, only two emitters are needed. The internal LED driver will drive the left emitter while a proximity reading is made, as shown in Figure 2. This value is converted to a digital count, a number between 0 and around 65,000, as the ADC in the VCNL4035X01 has 16-bit resolution.
The multiplexer will then switch from the output connected to the left emitter, now driving the right emitter while taking another proximity reading. Both readings are saved in separate data registers within the sensor, so that they can be read by a host microcontroller individually. The microcontroller will then acquire these measurements and compare the two readings.
If the hand is closer to the left emitter, the output count of the sensor will initially be higher for the left reading than for the right reading. As the hand continues with its right swipe, the output counts from the sensor will become approximately equal as the hand reaches the center between the two emitters. Eventually the output associated with the right emitter will be higher than the left as the hand completes the left-to-right swipe. This sequence of operations could be rotated ninety degrees to enable up-and-down gesture control.
The zooming function is enabled by measuring the output of the sensor as the hand is moved closer or further away from the sensor. The algorithms for detecting these gestures are predicated on a threshold proximity value which is exceeded when a hand is present: at this point, the sensor count should be higher than the offset value that the sensor reads when nothing is in front of it.
More robust and complex algorithms record the signal of each emitter in a continuous stream, and a measurement frame is overlaid to enable the shape of the signal to be analyzed. The percentage overlay of one emitter’s signal relative to the other can then be used to denote a time difference, which shows that a gesture has been made. Adding a third emitter will improve the resolution of the simple gestures described above and could possibly be used to recognize more complicated gestures.
There are many variables to take account of when implementing a gesture-recognition system:
- Size of objects other than a hand
- Distance from object to sensor and emitter
- Speed of the moving object
- Reflectivity of the moving object
- Distance between emitters
- Threshold levels required to avoid false detection events
- Ambient lighting and sources of optical interference
The use of discrete emitters gives the design engineer valuable flexibility, as their placement determines the size of the gesture and sensitivity zone. Some sensors integrate emitters within a single package, which removes this flexibility. The typical sensing range of an infrared emitter and sensor system is 20cm.
While the system shown in Figure 2 uses infrared emitters and light-to-digital proximity sensors, other systems offer millimeter granularity by combining a time-of-flight sensor with a 3D camera. This is a more expensive system, but it does allow the measurement of small finger gestures, and not only of large swiping motions. First-generation applications in vehicles today detect large motions, but second-generation systems under development will detect multiple finger gestures. For instance, the motion that a user’s fingers make when enlarging a photo on a smartphone screen may be recognised when made in the air. This requires the use of 3D camera systems.
Fig. 2: A two-emitter/single-sensor system can detect simple swiping and zoom gestures
This capability is similar to the motion-detection feature offered today by games consoles. For example, Microsoft’s Kinect system for the Xbox console detects motion at distances up to approximately 3m. Gaming systems track the entire body motion of the player, but their automotive counterparts will only need to track the driver’s hand gestures. The more complex the system for measuring gestures becomes, the more complex the middleware to convert the output of the sensor into a definable action.
Companies such as LG and Sony are joining more typical Tier 1 automotive manufacturers in developing automotive gesture-recognition systems. Customer comfort, multimedia, navigation and infotainment applications are taking on a higher priority as car makers implement more sophisticated Advanced Driver Assistance Systems (ADAS) and autonomous driving capabilities.
Eventually, there will be no need to avoid driver distraction: fully autonomous vehicles will have no driver, only passengers.