METHOD AND DEVICE FOR CALIBRATING PHYSICS ENGINE OF VIRTUAL WORLD SIMULATOR TO BE USED FOR LEARNING OF DEEP LEARNING-BASED DEVICE AND A LEARNING METHOD AND LEARNING DEVICE FOR REAL STATE NETWORK USED THEREFOR

摘要

The present invention relates to a method of calibrating a physics engine of a virtual world simulator used for learning a deep learning-based device, the calibration device comprising: (a) a virtual environment When the virtual current frame information corresponding to the virtual current state of the image is obtained, (i) by transmitting the virtual current frame information to the deep learning-based device, the virtual action information is output, and (ii) the virtual current frame By transmitting the information and the virtual action information to the physics engine, the virtual current frame information and the virtual next frame information corresponding to the virtual action information are output, (iii) the virtual current frame information and the virtual action information; outputting predicted actual next frame information by transmitting it to a real state network that has been trained to output predicted next frame information in response to an action on a real environment; and (b) calibrating a previously calibrated parameter to generate a current calibrated parameter.