Ambarella Inc., a developer of high-resolution video processing and computer vision semiconductors, and Momenta, an autonomous driving technology company, have developed a collaborative HD mapping platform for autonomous vehicles.

​The combined platform provides HD map technologies such as mapping, localization for autonomous vehicles, and map updates through crowdsourcing, thanks to the combination of Ambarella’s CV22AQ CVflow computer vision system-on-chip (SoC) with Momenta’s deep learning algorithms.

“We are pleased to partner with Momenta to provide a powerful and open HD map platform,” says Fermi Wang, CEO of Ambarella.

“The CV22AQ’s performance and advanced image processing help enable the full potential of Momenta’s advanced AI algorithms.” 

Highly scalable and production-ready, Momenta’s vision-based HD semantic mapping technology can create a closed feedback loop of big data, AI, and HD map updates through crowdsourcing. Momenta, which is based on localization, discovers changes in the map elements and provides frequent updates to the cloud.

Manufactured in an “advanced 10-nanometer process,” Ambarella’s CV22AQ provides the ultra-low power consumption that is needed for the design of compact automotive systems. Its CVflow architecture delivers real-time processing of up to eight megapixel resolution video at 30 frames per second (fps) for high-precision deep learning based object recognition, Ambarella says.

According to Ambarella, the CV22AQ’s high-performance image signal processor (ISP) delivers “superior image quality in low-light environments, while high dynamic range (HDR) processing extracts more image detail in high-contrast scenes,” which further enhances the system’s computer vision capabilities.

Using CV22AQ, Momenta can use a single monocular camera input to generate two separate video outputs. One is for vision sensing (perception of lanes, traffic signs, and other objects), while the other is for feature point extraction for self-localization and mapping (SLAM) and optical flow algorithms.

Ambarella notes that it provides a “complete set of tools” to help customers easily port their neural networks to the CV22AQ SoC. Based on the tool chain, Momenta can quickly migrate deep learning perception models to embedded platforms and achieve an accurate output.