In this research, we developed a 3D scanning system using multiple robots named CPS-VIII, which combines the laser measurement technique using multiple robots called CPS-SLAM and a high-precision 3D laser scanner. This system is able to acquire geometrical big data consisting of trillions of 3D points. The experimental results showed that the accuracy for CPS-VIII is 0.0085 % of the distance traveled, which means the error is 0.0231 m after the robot moved 270.1 m. In addition, we developed large-scale geometrical big data of urban area and surrounding area of Fukuoka city and provided them on the web. Furthermore, we developed space/road categorization techniques for autonomous vehicles and mobile robots, and proposed high precision recognition techniques using convolutional neural networks and the combination of multiple machine learning techniques.