Radars: Key Technology Trends Towards 4D Imaging Radars
Radars are a key element of the sensor suite in ADAS and autonomous mobility. The technology is already commercially used, particularly on various ADAS functions. Its use is set to increase in the short and long terms. In the short-term, legislation as well as voluntary safety commitments will push the adoption further. In the long term, the emergence of higher levels of autonomy will increase the radar content per vehicle, thus having a multiplier effect on the market.
These are exciting times for radar technology. Many changes are taking place. Here, we outline some key trends from our recent report "Radars 2020-2030: Technologies, Future Trends, Forecasts". This report provides a detailed assessment of the technology. It considers how radar technology is evolving. It examines the drivers and trends in frequency worldwide. It assesses the trends in the choice of the semiconductor technology, as well as the lithographic node. It analyses the trends in packaging and also the trends towards higher degrees of function integration within the chip, e.g., DSP. The report examines and critically benchmarks low insertion loss materials including ceramic-filled PTFE, PI/fluoropolymer, LCP, LTCC, and many other materials.
The report also considers how radar performance is improving, in particular with respect to higher azimuth and elevation resolution. The report analyses the developments on the signal processing side. It discusses how the point cloud of radars is becoming denser and how methods are emerging to allow radars to detect, classify, and track many objects in 3D space across all weather and light conditions. These efforts are at an early stage and can be expected to dramatically improve as larger manually or semi-automatically labelled training datasets become available and the fusion techniques evolve.
Finally, the report identifies and overviews the key innovative starts-ups and emerging products on the market. It offers an in-depth review of the ADAS and autonomous mobility markets. In particular, it offers market forecasts (2020-2040), in unit numbers, for passenger cars, robotaxis and trucks segmented by level of autonomy from 1 to 5. It then forecasts radar unit sales numbers. Lastly, it develops two cost scenarios for radar modules (SRR/MRR and LRR) and thus two market value projections. To learn more see "Radars 2020-2030: Technologies, Future Trends, Forecasts".
The semiconductor choice in automotive radars is evolving. First generation products deployed GaAs. They were at first used as bare dies mounted directly on boards and wire-bond connected. The next generation was on SiGe. This allowed higher on-chip integration of functions. The carrier mobility was high to allow high-frequency radars even at large lithographic nodes (e.g., 130nm). Advanced packages were also developed, evolving the technology from CoB to FOWLP or similar (fan-out-wafer-level-packaging).
Many are now developing Si CMOS (and SOI) technology. Many utilize the 40nm technology node, but some are pushing it down to 28nm or lower. The shorter channels support a high frequency even with a low carrier mobility. The small node, together with CMOS technology, allows higher integration of functions within the chip. Currently, the latest generation integrates not only the transceiver and the chirp generation, but also a microcontroller and digital signal processing (DSP) unit within the chip itself. This points the way towards single chip radar solutions. The switch to silicon technology will also better sustain a cost reduction path, especially as volumes expand.
Showing the evolution of semiconductor technology in automotive radars. The transition towards silicon-based technologies together with the advent of smaller lithographic nodes will mean that more capability can be on chip integrated, paving the way towards single-chip solutions able to support MIMO antennas.
The packaging and board technology have also evolved. First generation devices were composed of multiple dies mounted directly on the board (CoB) and connected by wire-bonding. These radar modules also had two separate boards: one for the RF and the other for digital functions. The packaging has evolved, and now various forms of wafer-level-packing are employed. The board also evolved to be hybrid with the top RF layer composed of a special low insertion loss material such as ceramic-filled PTFE or similar. In cases where a small antenna array will suffice, antenna-in-package (AiP) designs are already viable and some have been qualified for short-range automotive applications.
The choice of low insertion loss materials is critical and interesting. These materials will need to offer low loss tangent. Crucially, the dielectric constant and the loss tangent will need to remain stable against variations in temperature and frequency. Furthermore, moisture uptake will need to be low and the material will need to be easy- or with well-known modifications- processible, e.g., how to make the Cu stick. This study offers a comprehensive benchmarking of a wide range of materials on the market including ceramic-filled PTFEs, LCP, PI/fluoropolymers, ceramics such as LTCC or AlN, glass, etc. To learn more, see the IDTechEx report, "Radars 2020-2030: Technologies, Future Trends, Forecasts".
Benchmarking the loss tangent of various laminate and build-up materials. To learn more including about benchmarking of stability of loss tangent and the radar market size please visit www.IDTechEx.com/Radar
Towards 4D imaging radars
The capability of radar technology will rapidly expand. First, the antenna arrays are becoming larger. Some start-ups have designed and demonstrated radar chips able to support 6500 virtual channels. Such radar may reach 1deg azimuth and 2deg elevation resolution. This trend would enable rich 4D data points per frame to be obtained giving precise information on velocity, range, azimuth and elevation. Therefore, such radars would encroach into territory currently occupied by lidars although the latter will likely retain angular resolution and potentially object classification superiority. Higher data rates will pose interesting questions as to how much pre-processing will be required prior to sending the data out of the radar modules. It may, depending on the second fusion architecture, demand ultra-high speed (Gbps) links within the vehicle.
Furthermore, this and similar arrangements will enable the densification of the point cloud. This is key, because the sparse point cloud of data hinders advanced signal processing and will make data fusion techniques complex. These limitations have kept radars limited to detection of object presence as well as object velocity. They also make it difficult, at times, to distinguish between stationary and slow-moving objects.
The dense and high-resolution point clouds will enable the developments of object detection, classification, and tracking based on radar data. To achieve this, deep learning techniques are being developed. A major challenge though is the lack of extensive labelled training radar data. The manual labelling process requires expert input, and is thus expensive and time consuming. Now though companies are launching fleets for the capture of radar data in synch with camera, lidar, and other data, and are developing semi-automated labelling techniques which rely on late-stage fusion of data between camera, lidar, radar. Such efforts and techniques will accelerate the development of training datasets. Currently, such techniques are not as accurate as other means, but this is a fast-evolving space to closely watch.
Overall, this is one of the hottest trends. In the future, as radar technology shifts to smaller nodes, highly integrated packaged single solutions will emerge. The antenna array size will substantially expand, thus enabling better azimuth and elevation resolution, and thereby densifying the point cloud. The deep learning-based algorithms will also evolve in parallel, enabling radars to do object detection, classification, and tracking in 3D. In such cases, radars will begin to blur the boundaries with some lidar technologies whilst retaining the weather and light-level independence.
As mentioned above, this are exciting times for radars. We forecast that automotive radars can become a $12Bn market by 2030 in a moderate price erosion scenario. Note that the market will be, within that time frame, mainly pushed by ADAS. In the longer term (2030-2040), autonomous mobility (levels 3, 4, and 5) will drive the market. Here, the increase in radar content will counteract the emergence of peak car which we forecast to result from the rise of autonomous mobility.
To read more, please visit www.IDTechEx.com/Radar. This report provides a comprehensive view of the technology and market trends. It develops a comprehensive technology roadmap, examining the technology at the levels of materials, semiconductor technologies, packaging techniques, antenna array, and signal processing. It demonstrates how radar technology can evolve towards becoming a 4D imaging radar capable of providing a dense 4D point cloud that can enable object detection, classification, and tracking. The report examines the latest product innovations. It identifies and reviews promising start-ups worldwide. The report builds a short- and long-term forecast model covering the period between 2019 to 2040. The market- in unit numbers and value- is segmented by the level of autonomy and by passenger vehicles, shared vehicles, and trucks. In the first decade, ADAS (level 1 and 2) will be the primary market drivers whilst in the second decade autonomous vehicles will be.
These radar charts compare the status of today's radar with that which is emerging. To learn more about the technology trends that underpin this transformation please visit "Radars 2020-2030: Technologies, Future Trends, Forecasts".