UK-Japanese developer of Computer Vision Technology, Kudan Inc, has announced that its SLAM (Simultaneous Localization and Mapping) system is now ready for the market, with the technology suitable for a variety of applications, including autonomous vehicles.
Computer Vision is seen as the link between Artificial Intelligence (AI) and the Internet of Things (IoT), providing the ability for computers to visually map and interpret the world around them. Kudan has developed a real-time 3D tracking and mapping system that provides computers with a form of vision, allowing them to acquire, process, analyze and understand their environment and the objects within it, as well as their 3D position within that environment. The company has begun providing its SLAM technology to a number of industries, including self-driving cars, Advanced Driver Assistance Systems (ADAS), drones, robotics, and Augmented Reality (AR), where its precision tracking, high speed, low CPU usage and overall robustness can be a powerful tool.
Kudan has achieved this by developing a practical, next-generation algorithm, replacing current SLAM algorithms, such as ORB and PTAM., and taken them a step further and made a system ready for market consumption. KudanSLAM’s key features include:
Hardware Friendly – A flexible system that can be used with any camera or sensor, including monocular camera setups and rolling shutter cameras, and can also be embedded on processor architecture; High Speed / Low Consumption – Uses less than 5% of a mobile CPU, providing fast tracking, whilst maintaining good battery life; High Tracking Accuracy – Millimeter accuracy 1m away from the target; Robustness – Capable of operating even in extreme lighting conditions and can recover from sudden, jerky movements.
For self-driving cars and ADAS applications, KudanSLAM can be combined with internal sensors and lidar units to provide even more precise positional tracking at greater distances. It is also good for monitoring both the front and back of a vehicle accurately within a few centimeters, making it useful as a parking assistant. It can also provide navigation in areas where GPS cannot reach, such as in indoor spaces, underground or multistorey car parks, and ‘urban canyons’ or road tunnels. Kudan products are hardware agnostic and support most platforms from low-end through to the most advanced embedded applications, robotics and mobile devices.