Recogni, the developer of AI-based perception for autonomous vehicles, has announced Recogni Scorpio, which the company claims is “the world’s first 1,000 TOPS (Peta-Op) class inference solution for autonomous mobility.” did.
The company’s visual inference chip delivers superhuman object detection accuracy up to 300m in real-time under a variety of road and environmental conditions, processing multiple streams of ultra-high-resolution, very high frame-rate cameras. have the ability.
Microprocessor Report says Recogni’s solution “outperforms the inference engines of leading SoCs such as Nvidia’s Orin.”
RK Anand, Founder and Chief Product Officer of Recogni, said:
“From the beginning, we took a unique approach to processing high-resolution images at the edge to achieve near-perfect object detection and classification, enabling the autonomous driving stack to make better driving decisions.
“The Scorpio can handle 300 frames per second in less than 10 milliseconds using just 25 watts. We believe it will help accelerate implementation.”
Recogni’s solution is currently being evaluated by several leading automakers and suppliers and can achieve 1,000 TOPS with processing latency of less than 10 ms and power consumption of less than 25 watts.
Not only is it 10-20x more power efficient than competing solutions, it also enables flexible design of autonomous vehicle stacks with minimal impact on range.
In addition, the short processing time of less than 10 milliseconds gives the electronic control unit (ECU) enough time to make the necessary operational decisions. High computing power, efficient processing, low latency and low power consumption are the pillars of Recogni’s platform.
Marc Bolitho, CEO of Recogni, said: Add powerful new ADAS and autonomous driving capabilities to new vehicles.
“Recogni’s unique approach to recognition processing is truly first of its kind, enabling customers to deploy safe self-driving capabilities and extend the range of their electric vehicles.”
