Autonomous driving (AD) and advanced driver assistance systems (ADAS) rely on accurately sensing the environment around the vehicle to navigate safely. Global manufacturers have turned to advanced sensors and algorithms to improve perception and achieve unprecedented levels of safety.
Photo of a futuristic autonomous vehicle.
Edge sensor processing provider TERAKI has released the latest radar detection software that accurately identifies static and moving objects with greater accuracy and less computing power. The real traffic solution runs on ASIL-D compatible AURIX TC4x microcontrollers from Infineon Technologies AG.
“Automotive radar system performance has increased dramatically over the last few product generations,” said Marco Cassol, Director of Product Marketing at Infineon Automotive Microcontrollers. “Edge AI processing is one of many innovations that helped us drive this increase in radar performance. TERAKI’s unique radar algorithms are now being implemented in Infineon’s new parallel processing unit to demonstrate the next-generation radar performance of Infineon’s AURIX TC4x devices.”
“We have refined our algorithm to achieve more with less,” added Daniel Richart, CEO of TERAKI “With the minimum amount of data, our solutions correctly detect and classify static and moving objects with radar signals, providing AD and ADAS applications with the information essential for situational awareness and decision making. Ultimately, we aim to ensure security at the edge by reducing inference time and processing power required from constrained devices.”
As radar becomes the industry standard for cost-effective signal processing, overcoming the limitations of this sensor technology becomes a priority. For example, jamming can drastically reduce radar detection performance, leading to invalid detections in difficult multi-target situations, which also incur high processing requirements.
Furthermore, the accuracy required for reliable radar classifications involves more data points per frame and angular resolution of less than 1 degree for both static and moving objects to be correctly detected and classified.
TERAKI's machine learning (ML) approach aims to solve this challenge by working with raw data and reducing noise, while also acting as a cognitive function to dissect radar information, identify targets in a noisy environment, along with clusters and other interference, and decrease processing capacity at the edge. TERAKI's ML detection provides more points per object, leading to fewer false positives and greater security – especially when compared to other radar processing techniques such as CFAR.
Ported with Infineon's AURIX TC4x, TERAKI's ML-based algorithm reduces radar signals after the first Fast Fourier Transform, achieving up to 25x lower error rates for lost objects at the same RAM/fps. Compared to CFAR, classification is up to 20% higher in accuracy and valid detections increase to 15% more.
With this release, TERAKI is improving the chipset architecture of high-end devices, ensuring real-time processing performance on the AURIX TC4x, which alleviates computing requirements when consuming four or five bit bitrates instead of 8 or 32-bit, without compromising F1 scores. . This leads to a need for up to twice less memory.
