Forward camera systems involve high-speed video processing, complex sensor fusion, and real-time data analysis that enable the automobile’s corrective action. To achieve this high-level functionality with both mono and stereo camera systems, you may also integrate additional sensor types such as radar and laser sensors. Each sensor type is unique in how it provides data, making it a challenge to design for multiple architectures.

Design considerations:

  • High-performance parallel processing of video data and special algorithms for integration of multiple sensor data (radar, camera, etc.) at the lowest cost and with highest degree of integration
  • Increased bandwidth and video processing power at high resolutions and high frame rates to provide detailed, fast-moving images
  • No standardization in sensor types or algorithms
  • Many differing OEM design approaches
  • Low latency to provide real-time driver alerts and control of the vehicle

Traditional DSP processors or microcontrollers do not have enough power for real-time video processing and analytics and automotive system design needs additional hardware co-processors to keep up with these real-time processing requirements.

FPGA Integration

Instead of using DSP processors or microcontrollers, you can integrate the entire automotive vision system into a single, low-cost FPGA. For optimal system performance, you can develop the hardware parallel processing engines with FPGAs as well as integrate software algorithms running on our SoC’s hard processor system. In addition, FPGAs allow for customized I/O interfaces to various image sensors, along with output interfaces, such as CAN, LVDS, or Ethernet communications.

Implementation Options

The flexible nature of an FPGA gives you several implementation options. Altera's Cyclone V SoC FPGAs to integrate the radar, video, and sensor fusion processing algorithms along with network connectivity in a single device.

Forward-Looking Sensor Fusion ECU

With an FPGA system on a chip inside your engine control unit (ECU), you get:

  • Image processing intellectual property (IP) for video processing, sensor interfaces, and video analytics
  • Cyclone® V and Cyclone V SoCs for high-performance video processing engines and software algorithms
  • Automotive networking protocol IP for CAN, Ethernet, and LVDS interfaces
  • Scale image size and add overlays for parking assist using our Video and Image Processing (VIP) suite IP
  • High-resolution image sensor processing
  • Ability to scale performance across different FPGA densities in the same package footprint