The technology offers one of the most innovative solutions for highly advanced ADAS features such as Autonomous Valet Parking and Automatic Driving Assist
StradVision, a leading innovator in AI-based camera perception software for Advanced Driver Assistance Systems (ADAS) and Autonomous Vehicles (AV), is revealing its latest automotive Surround View Monitoring (SVM) technology at AutoSens 2020, an award-winning international conference for the automotive sensor industry.
Through an AutoSens 2020 online session titled ‘Discussion: How to run multi-channel cameras on a single automotive-grade SoC’, running from 19:45 to 20:15 GMT on November 19, StradVision will introduce perception software that automotive OEMs can add to its Surround View Monitoring (SVM) systems for Automatic Parking Assist (APA) and Autonomous Valet Parking (AVP) – both are highly advanced and fast-growing features in the ADAS technology field.
APA assists drivers to park a vehicle in a vacant parking lot, with or without driver’s intervention; while AVP provides a complete valet parking service controlled by ADAS – the vehicle drives itself to find a parking lot and returns to the driver once summoned.
As the market demand for APA and AVP increases rapidly, StradVision has been developing new technologies for more specific and precise recognition of parking areas. Its SVM technology can detect various static objects including poles, ground locks and stoppers, and road signs for path planning. It also integrates other cutting-edge technologies such as Visual Simultaneous Localization and Mapping (V-SLAM), Pseudo Lidar and depth estimation for accurate map generation, road profiling, and height classification.
“We have been testing our SVM solutions either through our own projects or in collaboration with external partners. Through this process, we are solving challenges in order to provide robust and stable performance, and provide strong compatibility with a wide range of system-on-chips (SoCs),” said Junhwan Kim, CEO of StradVision.
StradVision is advancing the future of autonomous vehicles through ‘SVNet’, an AI-based object recognition software that allows for ADAS and autonomous vehicles to detect and identify other vehicles, lanes, pedestrians, animals, free space, traffic signs, and lights, even in harsh weather conditions or poor lighting.
Compared to competitors, SVNet achieves much higher efficiency in memory usage and energy consumption, and can also be customized and optimized to any SoC thanks to its patented and cutting-edge Deep Neural Network-enabled software. SVNet works seamlessly with other sensors such as LiDAR and RADAR to achieve surround vision.
SVNet is currently used in mass production models of ADAS and autonomous driving vehicles that support safety function Levels 2 to 4, and will be deployed in more than 8.8 million vehicles worldwide.