Smart Inertial Sensors Are Critical for Safe and Sustainable AVs
In human mobility, we are facing a big dilemma on the road to safe and sustainable mobility. The pervasion of semiconductors in the car is increasing the apparent need for power. At a time when the world is striving to achieve energy sustainability, how do we square the growth in automotive semiconductors with sustainability?
Microelectromechanical systems (MEMS) and sensors have a critical role to play in ensuring the usefulness and safety of autonomous vehicles (AVs), and the latest developments are vital to achieve sustainability.
Taking mobility “onlife”
Let’s talk a bit about the evolution of MEMS and how the current trends in automotive connect with what we are calling the “onlife” era.
About 20 years ago, we started the “offline era.” At that time, MEMS technology could transform concepts into products. In the automotive world, MEMS inertial sensors enabled important innovations like airbags, which had a tremendous impact on passenger safety in vehicle collisions.
Over the past 10 years, we have moved from that offline era—where MEMS devices simply enabled certain product features or functions—to something much more powerful: the “online” era. Connecting sensors to the cloud has enabled performance improvements and technology fusion that was key to making their information available to any ecosystem. In our automotive example, this has enabled the transition from straightforward airbag deployment to also calling for emergency services and reporting pertinent information about the impact and vehicle position to assist responders.
This naturally brings us to the current “onlife” era. Here, there is no longer a difference between being online and offline. Instead, there is a basic fusion between technology and life. We’re talking about systems that are constantly vigilant and can sense, process and take action.
In mobility technologies, we now need a sustainable “onlife” era to deliver two major features. The first is human centeredness that improves the way we interact with the world around us. Human-centered technology is safe and noninvasive and acts as an extension of ourselves while enabling operating as a driver assistant. The second feature is sustainability to help us protect this incredible planet on which we all live.
So how do we move to the sustainable “onlife” era? We need to build sustainable sensorization of the world. Whereas in the past we had just a sensor (offline), then a connected sensor (online) and then a sensor that could sense, process and act (“onlife”), we now need self-configured sensors to optimize data-processing and ultra-low–power systems. This evolution is coming because we need to save the planet by reducing CO2 emissions to net-zero emissions by 2050.
Passenger cars contribute roughly 3 billion tons of CO2 emissions globally. Electrification has a central role in drastically reducing these emissions. At the same time, we’ll also see a significant increase in the adoption of vehicles with greater and greater levels of automation and, eventually, autonomy.
Inertial sensors for AVs: smart, safe, accurate
Smart inertial sensors are essential for higher levels of automated driving. These are Levels 3, 4 and 5, as defined by SAE International. They are also vital for complying with the stringent safety standards in place to protect vehicle passengers, pedestrians and other road users.
To fulfill these requirements, they must have three key attributes: They need to be smart, safe and accurate.
The smart attribute is essential: AVs need to be capable of reacting to every possible scenario. They must be driven by AI algorithms that can imitate (or improve upon) human behavior and reaction times. With processing integrated directly within the sensor, these can analyze the sensor data and execute operations without the latency and power requirements of vast sums of sensor data traveling to a host or cloud for processing. This speeds reaction time while dramatically reducing system power.
Sensors containing on-chip processing can take less than a few microamps to initiate the decision tree and less than 10 mA to analyze the readings. This is important because reducing the power needed to operate an AV or any large system is critical for sustainability. Consider, for example, a simple application like car monitoring. Smart sensors can monitor a car 24/7 with AI capabilities that allow the car to detect if it is being bumped, towed or vandalized when parked or can detect and adapt to environmental conditions or road surfaces when in motion.
Then appropriate programming of elements in an on-sensor controller can ensure the human-centricity of the sensors and allow the vehicles to offer self-configurable solutions with hardwired AI implementations, along with optimized power budgets that can contribute to sustainability goals.
Smart inertial sensors must also be safe. AVs must comply with the most stringent safety standards. Smart inertial sensors should enable AVs to have an accurate reading of their surroundings, as the car must know where it is, and where it is going, in relation to where every other vehicle is and is heading. And it isn’t just other vehicles; the AV has to know where every obstacle is, as that is also essential for safe operation.
Today’s vehicles are already integrating more and more embedded circuits implementing functional-safety systems in hardware for power efficiency. Inertial sensors have many roles, such as in compensating camera images affected by inclination and vibration caused by steering action and road noise. Certification to Automotive Safety Integrity Level B (ASIL-B) is commonly required for these types of systems. On the other hand, systems for automated driving face much more stringent requirements, calling for certification to, say, ASIL-D. The next generations of inertial sensors will be designed with this in mind, likely accompanied by independently tested software libraries to facilitate safety certification in accordance with ASIL-B and above.
In addition, the sensors must be robust, reliable and able to withstand harsh environmental conditions. They must also be secure to prevent unauthorized access and data breaches.
Lastly, smart inertial sensors need to be accurate. AVs require precise and accurate data to operate safely. We are asking these vehicles to drive for one hour with less than 0.1 degree of error and to initiate safety stopping that will bring the machine to a halt with an absolute accuracy of 20 cm. That’s incredible and demands accuracy comparable to that normally expected of a lunar guidance system. Yet we now need to achieve this with ordinary devices from our standard inertial sensing portfolio.
Ensuring the accuracy of the data helps minimize the application-processing workload and therefore power consumption by minimizing any need to polish the data.
Note also that latency affects accuracy: No matter how high the sensitivity, or how deep the resolution, the context changes continuously, so the information starts to become inaccurate as soon as it is generated. Low latency is a strong attribute of intelligent sensors.
The path to a fully autonomous vehicle is moving forward, but the destination is not yet visible. The major challenge, still, is sustainable autonomous driving. From the standpoint of sensor development, at least, we can say we have an idea.
In recent years, state-of-the-art MEMS has enabled ADAS applications that have demanded high-accuracy sensing capabilities, although very little to do with safety application development. On the other hand, the push among OEMs for advanced safety systems has stimulated rapid advancements in safety application development with limited need for accuracy in the sensing capability.
Now, if we really want to enable the path of sensorization for the future of a sustainable vehicle, we need to bridge the two needs: safety on one side and accuracy on the other. Let’s not forget that everything needs to be absolutely smart, with self-configurable sensors, with optimization of data processing in low-power systems. And this, ultimately, will enable carmakers to deliver the solutions the world needs.
There is no doubt that sustainable AVs are on the highway to our future. It is equally clear that smart, safe and accurate inertial sensors are critical. Like all automotive components, they must comply with stringent safety standards and withstand harsh environmental conditions and will help ensure sustainable AVs operate more safely than existing vehicles and with far greater efficiency. Sustainable MEMS technology is here to support this path.