東芝情報システム Official site

Middleware for self-position estimation necessary for mobile entities. It enables stable self-position estimation even in factory and indoor environments!

In AGVs and autonomous systems, it is necessary to accurately grasp the destination and self-position for route determination. Our company customizes Toshiba's unique self-position estimation technology for customer systems and optimizes it for real-time systems. ◆ Visual SLAM that operates with low resources This method estimates relative self-position based on the distance changes to multiple feature points visible in the camera. The algorithm, designed for real-time processing, achieves high-speed and low-memory operation. ◆ Improved accuracy and robustness through sensor fusion Using Visual SLAM or IMU alone presents challenges such as error accumulation during long-distance travel and the impact of environmental changes. By combining multiple sensors, our company addresses these issues, achieving high-precision and stable self-position estimation. ◆ Environmental recognition through surrounding map generation Based on environmental information, we can generate a spatial map of the surroundings in real-time while estimating self-position. This can be utilized for obstacle avoidance and error correction, making it applicable to harvesting robots and warehouse robots for picking tasks.