Michael Parker, Altera
EETimes (11/6/2012 5:02 PM EST)
Radar has long been used in military and commercial applications. Recently, radar has begun to appear on high-end automobiles for parking assistance and lane departure warning. Next-generation automotive radars must be more sophisticated, as radar will play a key role in active collision avoidance and adaptive cruise control. Once the radar is integrated into a system that actively controls the vehicle, as opposed to merely providing warning signals, the system reliability requirements become much more stringent, with an associated liability potential for vehicle collisions.
This white paper describes how an automotive radar system was built using digital processing segments with Altera’s rapid prototyping and development tool flow for digital signal processing (DSP) design, known as DSP Builder Advanced. The results provide actual circuit size and performance metrics for the digital portion of the radar processing. This digital processing incorporates a new breed of programmable logic, known as system-on-chip (SoC) FPGAs, which embed powerful dual 800-MHz ARM Cortex-A9 processors in a low-cost FPGA fabric. SoC FPGAs provide a flexible, scalable platform for applications from radar to light detection and ranging (LIDAR) to infrared and visible cameras. Moreover, both the FPGA hardware and ARM software implementations use floating-point processing, which provides superior performance in radar applications compared to traditional fixed-point implementations in FPGAs or ASICs. These low-cost SoC FPGA devices support volume applications, while providing much faster time to market compared to alternate ASIC solutions.
The SoC FPGA approach provides the ability for production line or even field firmware updates of both software and hardware, which can be important as automotive radar systems continue to increase in complexity. SoC FPGAs are also ideal for integrating video processing using cameras. Video analytics processing can be used in conjunction with the radar detection information in a process called sensor fusion, where multiple sensing systems combine to generate the most reliable data on which to base decisions.
Click here to read more ...