As the current state of technology stands, autonomous vehicles are not precisely “autonomous.” In fact, on the SAE levels of autonomy scale, Tesla’s cars are only at a 2 out of 5, which shows there is still room to improve.
The levels of driving autonomy. Image used courtesy of Synopsys
The result is that, even when the vehicle is in autonomous mode, drivers are still required to be fully alert and attentive, ready to take over the wheel at a moment’s notice. For this reason, many vehicle makers are beginning to develop and install driver monitoring systems (DMS) into their vehicles, which track things such as driver eye position and body language to determine whether or not the driver is focused on the road.
While Tesla already has some level of DMS in their vehicles, they’ve recently received approval from the FCC for a new mmWave sensor technology which they believe will increase safety in more ways than one.
DMS at a Lower Level
To achieve DMS functions generally requires the system to consist of several key hardware blocks. These often include a camera module, RF sensors, biometric sensors, and a dedicated processor (normally an SoC or ASIC). While the components are not necessarily unique, designing with them in an automotive environment can present many challenges.
A typical DMS block diagram. Image used courtesy of NXP
From a technological perspective, one of the biggest challenges in designing an accurate DMS is noise tolerance.
An automotive environment is rife with sources of interference such as the radio and the engine. When designing with RF sensors in a DMS, which are particularly prone to these kinds of interference, this can present a serious challenge.
Adding to that challenge is the need to consume as little power as possible to avoid draining the car battery. This constraint can be particularly challenging because these systems tend to run AI applications on the processor to perform the facial tracking that monitors the driver. Further, lower power electronics tend to be more noise sensitive, which is already an established problem.
From an economic perspective, designers need to make these systems as small and cheap as possible, hoping that there is space for them on the vehicle and that their inclusion won’t run up the overall price.
Tesla’s New Sensor Technology
Recently, Tesla finally received a new RF sensor technology that they claim is focused on passenger and vehicle safety monitoring.
In their Request for Waiver, Tesla explains that their new device is a mmWave radar sensor operating in the 60-64 GHz band. The device will consist of 4 TX and 3 RX antennas driven by a configurable radar front-end, which will modulate the radar signal to consist of consecutive frames.
These frames will include a transmit period in which a repetition of frequency chirps will be transmitted, followed by a listening period for RX, and finally, time for signal processing and idle time.
According to the technology description, the chirps will achieve a total duty cycle of 10%.
Tesla’s radar frames for their new mmWave sensor. Image used courtesy of Tesla
The sensor will have a maximum conducted power of +10 dBm, a maximum effective isotropic radiated power (EIRP) of +13 dBm, and will be focused on the vehicle interior, though it can also scan up to 2 meters outside the vehicle.
Tesla’s Plan and Future
Currently, Tesla states that their new sensor will monitor children left alone in a hot vehicle. They claim their sensor can detect things like breathing patterns and heart rates, helping make decisions about the child’s safety. The sensor will also provide depth perception and differentiate between a child and an object left on the seat to reduce false alarms.
Beyond this, Tesla hopes to use this technology to monitor a vehicle when parked, checking for damage or attempted entry.
Interested in other advancements towards “autonomous” driving? Catch up on the following articles below.