|Aglaia's Blackfin®-based "Electronic Eye" Warns Drivers of Lane Departure, Adjusts Speed|
For nearly a century, new drivers have been warned to "check the blind spot," the part of the road that cannot easily be seen in a car's mirrors. The fact is, safe driving means monitoring, analyzing and reacting to a huge amount of information about the car and its surroundings. In 2004, an estimated 8,116,000 crashes occurred in the United States alone, making the development of sophisticated driver-assistance technology an exceedingly high priority.
German-based Aglaia GmbH develops and markets special visual-sensor systems that assist automobile drivers, helping to prevent accidents and improve traffic flow. The company's real-time "electronic eye" solution is made up of hardware and software that mimics the human vision system from eyeball (image capture) to brain (image analysis and interpretation). Aglaia's system can notify drivers of upcoming traffic signs, provide warnings of potential traffic violations, automatically adapt speed to the speed limit, and warn drivers when they stray from the lane. The system can even provide "tiredness analysis" to help drivers avoid collisions.
Aglaia's complete solution includes cameras, image sensors, and image-processing algorithms. The traffic analysis application comprises software-configurable modules for functions like traffic-lane recognition, object recognition, traffic-sign recognition, and headlight and rear-light recognition. The application software, which runs on PC technology adapted for vehicle use, processes captured video images in real time.
At the heart of Aglaia's system are four Analog Devices, Inc. (ADI) Blackfin® ADSP-BF561 processors, each with two 600 MHz kernels. After a thorough market evaluation of off-the-shelf solutions, the company chose Blackfin because of its high performance, reasonable cost, low power, and versatile I/O options. Blackfin ADSP-BF561 processors enabled Aglaia to keep costs down and speed time to market, despite being in a non-commodity market where economies of scale are not a significant factor.
Aglaia's developments are consistent with the continuing trend toward the use of electronic components in automobiles to enhance driving comfort and, above all, improve safety and prevent accidents. The company believes the next technological step is the use of camera-based visual sensor systems. In fact, some applications, such as reversing cameras, night vision, lane departure warnings, and traffic sign recognition, cannot be realized without visual-sensor technology.
Most of the processing is handled by PC technology adapted for in-vehicle use by a "technical platform," or "TEPLA," Aglaia's low-cost electronic control unit (ECU). The unit houses four ADI Blackfin ADSP-BF561 processors, offering a total of 4.8 GHz of processing power, which can be extended to up to eight processors per board. Plugging two TEPLAs together doubles the processing power.
Aglaia's system comprises a number of products, including rugged CMOS cameras developed specifically for road vehicles. The cameras interface to Aglaia's TEPLA via a FireWire connection. The TEPLA in turn can interface to the vehicle's controller area network (CAN)-Bus, which aggregates data from the vehicle's system of sensors. This is how data is acquired and recorded for all of the driver assistance systems. Image signals are transmitted digitally via the local interconnect network (LIN)-Bus. MobileScope, another product that is part of Aglaia's system, records, interprets, and reproduces video data from the cameras, plus sensor data, and CAN data. Aglaia also provides a flatscreen that can be attached to the headrest of the front seat of a vehicle, and an industrial-sized keyboard and trackball.
Built for Speed
Aglaia chose Blackfin because it offers extremely high performance and low power consumption. The company needed to consume less than 20 Watts at about 3 amps – plus a combination of embedded micro-controller and signal processing – all in one chip. Each of the three dual-core 600MHz BF561 devices used in the 1.2V Aglaia design draws less than 1A, easily beating Aglaia's 20 Watt limit.
The ADSP-BF561 processor is a high-performance member of the Blackfin family of products, with two independent ADI processors, each with a dual-MAC signal processing engine, a RISC-like instruction set, and SIMD multimedia capabilities in a single instruction set architecture. The Blackfin processor's unique combination of processing attributes eliminates the need for separate digital signal and control processors, which reduces bill of material (BOM) costs and greatly simplifies hardware and software design tasks. Blackfin processors also provide world-class power management, featuring a low-power design (at 3 amps Blackfin consumes just over 10 Watts, worst case) and low-voltage design (Blackfin can be powered by an external supply as low as 2V) that lowers the overall power dissipation.
The four Blackfin ADSP-BF561 processors inside Aglaia's TEPLA will crunch on low-level algorithms, like edge detection routines, that are easily implemented since their repetitive nature can be efficiently synthesized. The processors will also crunch on higher-level algorithms that require all the power a high-performance signal processor can apply to the problem. Real-time algorithms, for example, must compute the orientation and the lateral pose of a vehicle with respect to road and lane-marking features "observed" by the cameras. Curve detector algorithms that automatically handle occlusion by vehicles, signs, light spots or shadows, or low-image contrast are complex and involve the kind of mathematically intensive processing only achievable on a high-performance signal processor such as Blackfin. Likewise, algorithms that correlate data from two cameras via triangulation to detect objects located above the roadway and/or relative to the vehicle in 3D space also require a high-performance signal processor. Blackfin will offer scalable processing power for image processing and other tasks, such as communications between the TEPLA and the network of cameras and sensors that interact with the vehicle's CAN-Bus and LIN-Bus.
Another important reason Aglaia chose Blackfin was for its I/O and memory options, including its parallel data interface (PPIs) and direct memory access (DMA) features. Blackfin offered a large amount of on-chip memory and the I/O options that the company needed for frame grabbing and transferring data at extremely fast speeds. Blackfin also operates within the recommended temperature range. It is the first processor Aglaia saw at this price point that was suitable for use in automotive applications.
Aglaia's TEPLA includes six IEEE 1394b FireWire ports for three independent busses for the attachment of cameras and other equipment, three CAN-Bus interfaces and one LIN-Bus interface for direct automotive data processing. TEPLA also includes one Ethernet interface to allow two TEPLA modules to be plugged together, and an optional USB interface. The system features a rugged design and operates on low power.
Aglaia used ADI's Visual DSP++, an easy-to-install and easy-to-use integrated software development and debugging environment (IDDE), which enabled efficient management of the project from start to finish. ADI provided a very high-quality tool chain with its compiler, linker, and debugger. Aglaia also used ADI's PCI-based JTAG emulator and interface to bring up the application quickly.
Due to the modular nature of the company's products, Aglaia's image processing solutions can be used in other application areas besides traffic flow analysis. The technology is well suited for security and biometrics. The software also includes modules for color image analysis, which is often used in microbiological applications. Because of the cost/performance advantages of Blackfin, Aglaia predicts that many new applications will be ported to TEPLA in the future. In turn, Aglaia will be on the lookout for new processors from ADI that sport more internal memory, more I/O options, and even higher processing speeds.
For more information, visit Aglaia's website.
Read more Customer Case Studies like this...