Ceva recently announced that Nextchip has licensed the NeuPro-M Edge AI Neural Processing Unit (NPU) IP for its next-generation advanced driver assistance systems (ADAS) solutions. Nextchip develops ADAS and Image Signal Processors (ISP) solutions for automotive Tier 1s and OEMs to design high-quality viewing cameras that deliver exceptional performance in any lighting condition and weather scenario.
According to data from Grand View Research, the global ADAS market is expected to reach $122.8 billion by 2030, growing at a CAGR of 19.4% from 2023 to 2030, driven by increasing demand for safety features in vehicles and autonomous driving. To further enhance its next-generation ADAS solutions, Nextchip has selected the Ceva NeuPro-M NPU IP, to leverage its advanced AI and vision capabilities to unlock unparalleled levels of performance, efficiency and accuracy.
In particular, the NeuPro-M’s robust support for Vision Transformers (ViTs) brings superior performance for ADAS vision systems, including object recognition, segmentation and Free Space detection in complex scenes, such as in cluttered environments (e.g., a pedestrian partially occluded by a parked car). Furthermore, with multiple processing engines, Ceva NeuPro-M enables the processing of multiple video streams and AI models in parallel for higher efficiency and precision.
“The ADAS market is constantly evolving, with new innovations in AI like vision transformers helping to deliver major improvements in performance and safety. By integrating Ceva’s NeuPro-M Edge AI NPU into our automotive safety solutions, we can leverage these new innovations to enhance the accuracy and efficiency of our ADAS systems, ultimately enabling safer and more reliable vehicles on the road,” said Ewoo Chon, CTO of Nextchip.
“Our NeuPro-M Edge AI NPU enables ADAS system designers to push the boundaries of AI performance in the most advanced vision-related ADAS applications. Nextchip’s next generation ADAS solutions featuring the NeuPro-M will help automakers and Tier-1s add more robust ADAS capabilities, enhance system reliability, and contribute to a safer driving experience for consumers,” added Ran Snir, General Manager and Vice President of the Vision Business Unit, Ceva.
The Ceva-NeuPro-M is a high-performance, low-power Edge AI NPU architecture engineered to accelerate deep learning and machine learning workloads in a wide range of applications. With its scalable architecture and advanced processing capabilities, the NeuPro-M enables efficient processing of complex neural networks, including convolutional neural networks (CNNs), recurrent neural networks (RNNs), and vision transformers (ViT).
With a scalable processing range of 4 to 400 TOPs per core and leading area efficiency, the Ceva-NeuPro-M optimises key AI models seamlessly. Its support for various fixed-point and floating-point data types and formats ensures highest efficiency of neural networks realisation, and with its built-in DSP (Digital Signal Processing) capabilities it makes a well suited solution for applications requiring high-performance AI processing, such as advanced driver assistance systems (ADAS), computer vision, and more.
There’s plenty of other editorial on our sister site, Electronic Specifier! Or you can always join in the conversation by commenting below or visiting our LinkedIn page.