22 Oct 2024
Leading in Safety: How OEMs Can Meet NHTSA's Stringent AEB Requirements While Controlling Costs
Similar Posts
22 Oct 2024
In April 2024, the National Highway Traffic Safety Administration (NHTSA) finalized stricter regulations for Automatic Emergency Braking (AEB) systems. By 2029, all new cars sold in the U.S. must meet these advanced AEB standards, raising the bar from the current voluntary commitments made by original equipment manufacturers (OEMs).
These updated rules require AEB systems to function at higher speeds of up to 90 mph, and they must also work effectively during nighttime driving using only low-beam headlights. Additionally, a new "no contact" performance requirement demands that AEB systems fully prevent collisions when a vehicle is traveling up to 62 mph for cars and 45 mph for pedestrians.
OEMs selling cars in the U.S. will need to pass a battery of tests to demonstrate compliance, including nighttime assessments and trials with stationary objects simulating cars and pedestrians.
In preliminary NHTSA tests, only one out of 13 2023 models passed, but even then, the vehicle engaged harsh braking—uncomfortable for passengers and potentially dangerous at higher speeds.
The Insurance Institute for Highway Safety (IIHS) also conducts AEB testing and has updated their test scenarios to reflect the new regulation requirements. Since June 2023, the organization has rated 194 model year 2023 AEB systems tested during the day. Only 33 models (17%) fully avoided the pedestrian mannequin in every test condition. Of the 114 model year 2023 PAEB systems tested at night, only 12 (11%) fully avoided the pedestrian mannequin in every test condition.
Clearly, most vehicles will require upgrades to meet these regulations by the 2029 deadline.
According to NHTSA, the new regulations are expected to save approximately 360 lives annually, prevent 24,000 injuries, and reduce the economic toll of collisions. Yet, the automotive industry is lobbying against the new regulations.
The Alliance of Automotive Innovators, the automotive industry’s main lobby group, asserts that as they stand the new regulations are not practicable and that NHTSA failed to consider adequately the costs of the new requirements, including the “likely disbenefits” that will be induced by the new standard.
Initially NHTSA estimated that a software upgrade would be enough to comply with the new standards and that total costs across the industry would not exceed $282 million. However, OEMs argued that the changes would require additional hardware, leading NHTSA to revise its estimates upward to an annual cost of $354 million.
The updated estimates still do not come close to the $430 million per OEM that the Alliance estimated as the cost to meet the new standards, which would place the total cost in the billions of dollars. In a letter to the NHTSA, Bosch also pointed out that certain vehicle models “may require significant hardware updates, such as the inclusion of more advanced sensors (e.g. radar, cameras), increased computing power, and/or improved brake systems to meet the specified requirements.”
The Alliance also stated that the performance requirement for “no contact” with the lead vehicle or test mannequin at high speeds will lead to unsafe unintended consequences, namely an increase in the number of false positive AEB engagements and a resulting increase in rear-end collisions. Mitsubishi, and Toyota echoed the Alliance’s concern, and added that in some situations AEB activation while traveling at high speed may induce unstable vehicle dynamics. The Alliance also requested that the stationary pedestrian test not be required at night, highlighting the challenge of low illumination for current ADAS.
This leaves OEMs with the pressing challenge: how can they develop a compliant AEB system that is both reliable to the performance standards and affordable?
In this article, we explore the options automakers are considering.
OEMs must first choose which sensors to integrate to meet the heightened performance requirements, balancing cost and reliability.
Cameras are the most widely used sensors in ADAS (Advanced Driver Assistance Systems) and are already included in almost all new vehicles. Cameras have limited range, however, and their performance drops significantly at night. The testing conducted by NHTSA and IIHS shows significant performance degradation of systems when restricted to low beams during night driving and when a pedestrian enters the roadway from an occluded position.
Most current vehicles use inexpensive, low-resolution cameras primarily for lane-keeping with only higher end models using higher resolution cameras for more advanced ADAS functions. For high-end cars, a software upgrade may be sufficient to meet regulations while the rest of models would need significant hardware upgrades. In their commentary to NHTSA, Honda stated that camera resolution is limited by the pixel count on the image capture chip and that at longer distances, the number of pixels for an object will be reduced, resulting in blur that makes it difficult to detect objects. Further, Honda stated that a higher resolution can be achieved only through new sensor hardware that would require further developmental work as well as more processing power, including a change of imaging processing electronic control unit (ECU). These upgrades are expected to add $200-$300per vehicle.
Moreover, even with all these upgrades, it is unclear whether cameras alone can provide reliable perception in all cases, such as night driving with low beams, inclement weather, and situations where a hazard is occluded and revealed with too little time for the vehicle to respond.
Lidar offers high-resolution imaging, but it comes with a steep price tag. While it can improve nighttime performance, lidar shares some of the same weaknesses as cameras, such as limited range and limited effectiveness in heavy rain and fog.
Lidar is likely to be cost-prohibitive for most mass-market vehicles, with estimates ranging from $500 - $1,000 per vehicle. Even for high-end models, lidar alone may not offer enough reliability without additional sensor support.
Infrared cameras have strong detection capabilities, especially for identifying pedestrians at long ranges, regardless of lighting conditions. However, no vehicles today use infrared cameras for collision avoidance, and adding them would require both new sensors and processing hardware.
The total system cost for infrared solutions is expected to be around $300 per car, and infrared still will not offer improvement in vehicle AEB at high speeds due to limited range. Therefore this approach would require other sensors or sensor upgrades for a complete system solution.
Radar is already widely used for AEB, Forward Collision Warning (FCW), and Adaptive Cruise Control (ACC). It is known for its long-range capability and reliability in all weather conditions. When NHTSA revised cost estimates for the new rules, the stated purpose for the increase was for installing a radar sensor on the 5% of cars without one today, indicating they believe radar is essential for meeting the new performance bar. Traditional automotive radars, however, have limited resolution and struggle to detect stationary objects, which is crucial for the new AEB standards.
Emerging technologies, like cascaded imaging radar and Massive-MIMO radar (offered by companies like Mobileye and Arbe), improve radar resolution, making them more capable of detecting stationary hazards. This advanced radar technology comes with higher costs, however, pushing OEMs to look for more affordable alternatives. A state of the art cascaded radar like a 4-chip imaging radar adds $125 - $175 in cost while a Massive MIMO imaging radar is substantially more expensive.
One low-cost solution comes from Zendar, whose AI radar perception software brings critical capabilities like stationary object detection and low-latency object classification to the existing radar, which is standard on 95% of cars in the United States. OEM’s can enhance the performance of the existing radar through software solutions provided by Zendar and meet the AEB requirements without hardware changes, minimizing cost.
For most OEMs, a radar-camera combination will offer the best balance between performance and cost. Both sensors are already installed in most new vehicles, and upgrading them would be more economical than adding entirely new systems. Each sensor has complementary strengths. Cameras are most effective during daytime driving, while radar offers long-range, all-weather reliability.
Upgrades to camera sensor hardware would be beneficial but may not be enough for OEMs to become compliant to the new regulations. Radars in production today have critical limitations which limit their utility for safety functions like AEB/PAEB. If a software upgrade for radar perception is able to overcome current limitations, it will be easier and more cost-effective to meet the new regulations without major modifications to the camera system.
A smart solution will effectively leverage the strengths of both radar and camera systems to deliver comprehensive, reliable ADAS functionality. Cameras excel in lane detection, traffic sign recognition, and short-to-medium-range object recognition—insofar as lighting and weather conditions permit. However, using cameras for more advanced tasks like long-range detection or velocity estimation puts substantial demand on both the sensor and the processing unit, requiring more advanced hardware and software. High-resolution cameras and powerful image-processing ECUs would be needed to meet the new NHTSA standards, driving up costs significantly.
A smart radar-camera fusion model provides a more cost-effective alternative. Radar, especially when enhanced by AI-driven solutions like Zendar’s, can reliably detect stationary objects and analyze complex scenes quickly, offering more consistent performance than existing low-resolution cameras in scenarios like high-speed driving and low-light conditions. This allows radar to take on the bulk of the workload in areas where it excels, leaving cameras to focus on tasks that suit their strengths.
Radar’s reliability, especially with AI upgrades that improve its perception capabilities, allows manufacturers to enhance vehicle safety and meet the new regulations at a low cost. Instead of "beefing up" camera systems to compensate for their weaknesses, OEMs can deploy a complementary fusion model, where radar delivers superior performance in critical scenarios which allows more affordable, incremental improvements to the camera system.
The Alliance for Automotive Innovation has stated that their goal is not to oppose NHTSA's decision to regulate AEB and PAEB. Their primary concern is the stringent performance requirements, which they argue could drive up costs and introduce safety risks, such as false AEB activations at high speeds. In contrast, NHTSA maintains that the new standard maximizes social benefit, noting that some vehicles are already meeting the standard today, making the bar both achievable and a necessary step for improving road safety.
With the new regulations set to take full effect by 2029, OEMs face a critical time frame for innovation. Although the five-year window may seem ample, the complexity of upgrading or redesigning AEB systems while maintaining affordability demands immediate action. Fortunately, technologies such as AI-enhanced radar perception offer a viable, cost-effective path for many manufacturers, allowing them to meet safety goals without excessive hardware costs.
The journey toward a collision-free future comes with its challenges, but the solutions are well within reach. OEMs that begin exploring and adopting advanced, cost-effective technologies—like AI-first radar perception—will not only be prepared to meet the 2029 deadline, but will also be at the forefront of creating safer roads for everyone. Now is the time to take proactive steps, embrace innovation, and lead the way in advancing vehicle safety technology.