So Andy speak about velodyne accuracy which is +- 2 cm so he speak for 4 cm accuracy. But this is the accuracy of the sensor. Which is the accuracy of the registered model? I have seen other solutions which use gps and imu for the registration and promise 15 cm absolute accuracy with the same velodyne sensor. I understand that Andy use a Slam method which sound perfect but what is the real accuracy is what matter in the end. With photogrammetry we can archive 1 pixel horizontally and 2 verticak and if Good contrlop points measured this will be the accuracy of the final model.