Sensor Fusion Is an Enabler in Automotive ADAS Applications
October 18, 2023
From the exterior, cars of today don’t look that much different from those manufactured a decade ago. However, if you peeled back the skin and looked inside, you’d see huge changes.
Those changes would be under the hood in the engine control and management, in the infotainment system, and in the ADAS (advanced driver assistance systems) feature set.
This latter segment has taken on an increased importance as it has become clear that the number of car accidents and serious injuries can be reduced through the use of these automated systems. Research shows that the majority of accidents are caused by human error. To combat these safety issues, the latest ADAS tools help vehicles stay in their lanes, alert drivers when another vehicle has entered its “blind spot,” and can automatically apply the brakes if necessary.
Another shift taking place in automobile design is the move from individual electronic control units (ECUs) to one central ECU. Previously, it was not uncommon for each function, whether it’s the anti-lock brakes, the rain sensors, or the radars to detect vehicle spacing, to each have its own ECU. That was clearly overkill and made for a more complex and costly design. Fewer ECUs means less wiring, which reduces the vehicle’s weight.
Performing upgrades to a central ECU is far simpler as well, as there is just one, albeit much more complex, code base to update. Having a singular code base should lower development costs and reduce bugs. And a byproduct of the central ECU, from the OEM’s perspective, is that it provides a potential revenue stream in the form of upgrades. The central ECU also reduces the number of entry points, thereby increasing the vehicle’s security.
Note that some of the legislation for these features comes from the European New Car Assessment Program (Euro NCAP). While they are not producing mandates, the group has created a five-star safety rating system that’s used to compare vehicles and identify their various safety characteristics. The ratings are determined from a series of vehicle tests that represent, in a simplified way, real-life accident scenarios that could result in injured or killed car occupants or other road users.
Data Pulled From Sensors
Sensor fusion, the process of integrating data from multiple sensors to create a more comprehensive and accurate representation of the vehicle's surroundings, is simplified by the use of a central ECU, as all the data leads to the same destination. Sensors that could be “fused” include cameras, lidar, radar, GPS, and IMUs (inertial measurement units), all which are used to gather information about the environment and the car's own state.
By adding processing capabilities to those sensors, we get today’s ADAS features, like those mentioned above. And as the number of cameras (and their resolution) and other sensors increase, the amount of data that’s collected and needs to be processed grows at an astounding rate.
Those cameras are used externally for object recognition and reading road signs, while lidar helps create a three-dimensional map of the surroundings. Radar can detect the speed and distance of objects. While each only represents one piece of the puzzle, when combined, they form a complete solution. Internally, the cameras can tell the state of the driver and backseat passengers.
Once the data is collected, it’s transferred to the car’s AI accelerators, which run a series of neural network algorithms. Those processors consider various elements, including the environment, the car’s speed, the weather conditions, and so on, to make a decision. The latest AI processors and associated algorithms are becoming more sophisticated, enabling vehicles to learn from real-world data and improve their driving capabilities over time.
Another trend that goes hand-in-hand with the adoption of a centralized ECU is the use of multi-Gigabit Ethernet type of networks. The higher bandwidth is needed as more data is coming to the central location.
Infineon, a vendor with a long history in the automotive space, is a one-stop shop for many of these technologies and associated components. In fact, its products ensure that the system/vehicle remains safe, responsive, and reliable. For example, Infineon offers a wide variety of automotive radar sensors including short-, mid-, and long-range sensing.
The XENSIV family of sensors consists of the RASIC 77/79-GHz front-end radar sensor ICs that can detect and recognize objects at a range of up to 250 meters. The parts support ASIL C, which reduces the design effort and also contributes to obtaining the desired five-star Euro NCAP rating. The 77-GHz parts represent the current generation, while the next generation is on the horizon. That newer version will extend the part’s capabilities, particularly the range, which will go beyond 250 m.
Infineon’s highly integrated 24-GHz radar transceiver family can reduce board area by around 30% when compared to discrete offerings. These sensors are intended for RF front-end modules for mid- and short-range automotive radar applications.
Between the two is a 60-GHz radar transceiver that the company claims is the most power- and cost-effective single-chip solution for in-cabin monitoring systems. It is used for rear-seat child detection, which is being mandated in certain parts of the world.
Another product that contributes to the ADAS functionality is Infineon’s high-performance MEMS microphones, which are also part of its XENSIV sensor family. The parts are qualified according to the AEC-Q103-003 automotive quality standard.
The parts enable distortion-free audio capture for all speech-related activities in the car, improving speech intelligibility for voice recognition algorithms. They offer acoustic noise cancellation (ANC) with flat frequency and stable phase response. Other highlights include close sensitivity and phase matching, suiting the mics for beamforming arrays. They can also contribute to road safety with the ability to pick up sounds like sirens from emergency vehicles or detect dangerous road conditions.
In conclusion, it takes a village of components to create and deploy automobiles with the latest ADAS. Features. Infineon has many of those bases covered.
Tim Grai, Principal Engineer, Vehicle Automation and Chassis Applications, Infineon Technologies. Tim has over 30 years of experience in the automotive industry in technical, marketing and leadership roles at OEMs, Tier 1s and semiconductor suppliers. He has a special interest in embedded software development and has led development projects ranging from powertrain controllers, to electric power steering, body controllers and instrument clusters.
Priya Muralidharan, Director, Vehicle Automation and Chassis Marketing, Infineon Technologies. Priya comes with 15+ years of experience in automotive having worked with Tier 1s and semiconductor suppliers in the areas of embedded software and business development for ADAS and chassis applications.