Thanks to AI algorithm upgrades, more and more car companies are adopting vision-based perception solutions. This solution abandons light detection and ranging (LiDAR) technology in favor of cameras and millimeter wave-radars, which significantly reduces the system’s hardware cost. However, pure vision solutions require advanced algorithm capabilities and massive amounts of data for training, and their detection accuracy is susceptible to weather and environmental factors. For autonomous driving, device redundancy is a crucial safety requirement. Therefore, the multi-sensor fusion perception scheme remains the preferred choice of most car companies at this stage.
As the most important sensor for autonomous driving, the resolution of cameras are continually upgraded, with the number of 8MP cameras on board rapidly increasing. Compared to 1-2MP cameras, an 8MP vehicle camera offers longer detection distance, wider field of view, and higher image quality. The current highway NOA systems are mainly adopting 8MP cameras. With increasing attention being paid to the 360 ° Around View Monitoring (AVM) System, the use of surround view cameras has significantly increased. High-level autonomous driving systems prefer 11V or 12V schemes (2-3 front view cameras, 4 side view cameras, 4 surround view cameras, and 1 rear view camera), leading to increased installations of front and side view cameras in vehicles.
LiDAR is a core sensor for high-level autonomous driving systems, as it outperforms other sensors in terms of detection range, measurement accuracy, and anti-interference capabilities. Currently, the main obstacle to the large-scale application of LiDAR comes from its high cost, following the pressure faced by the industry to reduce cost last year. Price wars between car manufacturers further hinder the popularization of LiDAR in vehicles. Product innovation is a key approach to reducing the cost of LiDAR to an acceptable range. The evolution from mechanical to solid-state LiDAR, as well as the use of semiconductor chips to replace the traditional discrete architecture, are current research and development directions of LiDAR manufacturers.
Meanwhile, 4D millimeter-wave radar has gained a lot of attention in the autonomous driving community in recent years. Conventional 3D millimeter-wave radar has shortcomings, which include a short detection range, poor precision, and an inability to determine an object's height. These issues will be effectively resolved by 4D millimeter-wave radar. By combining algorithms, 4D millimeter-wave radar will even be able to identify objects, making it adaptable to more complex road conditions. However, 4D millimeter-wave radar is still unable to compete with LiDAR in terms of resolution. It may take the role of low-resolution LiDAR in some mid-high level autonomous driving systems in the future. Blind spot detection will be an important application of 4D millimeter-wave radar.
Omdia has been conducting in-depth research and study on emerging technologies and trends in automated driving perception systems. The related data and insights are detailed in our latest report, "Automotive CMOS Image Sensor & Camera Module Report - 2023".
For more insights and analysis covering market trends and industry forecasts prepared by Omdia’s Semiconductors practice, click here.
More insights
Assess the marketplace with our extensive insights collection.
More insightsHear from analysts
When you partner with Omdia, you gain access to our highly rated Ask An Analyst service.
Hear from analystsOmdia Newsroom
Read the latest press releases from Omdia.
Omdia NewsroomSolutions
Leverage unique access to market leading analysts and profit from their deep industry expertise.
Solutions