Japanese automotive supplier Denso is using AMD’s adaptive computing technology in its next-generation lidar platforms. This enables the platform to achieve a more than 20-fold increase in resolution at low latency, providing greater accuracy in detecting pedestrians, vehicles, clear space and other elements.
Denso’s lidar platform, scheduled for launch in 2025, will use the AMD Xilinx Automotive (XA) Zynq UltraScale+ adaptive SoC and its functional safety suite of developer tools to achieve ISO 26262 ASIL-B certification. The XA Zynq UltraScale+ family integrates multiple, differently optimised processing units as a multiprocessor system-on-a-chip (MPSoc) in one platform. Denso uses these MPSoCs in its Single-Photon Avalanche Diode (SPAD) lidar system, which claims to produce the highest point cloud density of any lidar system currently on the market. Point cloud density describes the number of points in a given area and is comparable to image resolution in optical systems. The higher the resolution, the greater the probability that decision-relevant details will be captured.
In general, SPAD-based systems are used by car manufacturers because of the space savings they can achieve. The highly adaptable XA Zynq UltraScale+ MPSoC enables Denso’s lidar systems to reduce the size of current lidar implementations, allowing multiple such sensors to work together for both the forward and side views of a vehicle. One device can be used for multiple Denso lidar systems. The adaptability of the Zynq UltraScale+ also enables its use for future generations of Denso lidar. This lowers system costs and helps to future-proof designs.
- Continental relies on Xilinx SoCs for next-gen radar systems
- Subaru uses Xilinx chips in central driver assistance system
- Xilinx bolsters MPSoC family for more computing power
Today’s production vehicles typically have at most one forward-facing lidar, while next-generation vehicles will usually have multiple systems, including forward, rearward and side-facing lidars. The additional systems are needed to enable full autonomy beyond driver assistance. Denso Lidar can also be used for infrastructure monitoring, factory automation and other non-automated driving applications.
Denso’s SPAD Lidar can generate over three million points per second at 10 frames per second. This mid-range lidar system uses XA Zynq UltraScale+ MPSoCs for the system monitoring function to ensure correct temperature and overall system operation. Because the system uses time-to-digital conversion instead of analogue-to-digital converters, the size and cost of the overall system can be optimised while still delivering high-performance and high-density data.
- Farewell Xilinx as AMD deal completes
- Denso packs radar and vision in global safety product
- Denso, Toyota launch semiconductor JV
As the pace of innovation in the automotive industry continues to accelerate, the need for high-performance computing power, compute acceleration and graphics technologies is increasing. AMD aims to meet these demands by offering a wide range of high-performance CPUs, GPUs, FPGAs and adaptive SoCs. From powering in-vehicle infotainment systems to ADAS, autonomous driving and networking applications where functional safety is paramount, AMD aims to provide automotive manufacturers with a one-stop shop for silicon and software solutions. To achieve such a broad positioning, the chipmaker acquired competitor Xilinx, which had developed the Zynq UltraScale+ MPSoCs, just under a year ago.