METHOD AND DEVICE FOR OPTIMIZED RESOURCE ALLOCATION IN AUTONOMOUS DRIVING ON THE BASIS OF REINFORCEMENT LEARNING USING DATA FROM LIDAR RADAR AND CAMERA SENSOR

摘要

The present invention relates to a method and apparatus for optimized resource allocation for use in autonomous vehicles. More specifically, in a method for efficiently performing resource allocation during autonomous driving by reinforcement learning, (a) when a computing device acquires at least one attention sensor data, an attention network causes the attention sensor data with reference to calculating one or more attention scores by performing at least one neural network operation; (b) the computing device causes the detection network to acquire, with reference to the attention score, at least one video data acquired by at least some of the one or more cameras installed in the autonomous vehicle, the video data with reference to generating at least one autonomous driving decision data; (c) the computing device causes the drive network to operate the autonomous vehicle with reference to the determination data, and obtains at least one situation data indicating a change in a surrounding situation of the autonomous vehicle in operation and generating at least one reward with reference to the context data; (d) causing, by the computing device, the attention network to adjust at least a portion of one or more parameters used in the neural network operation with reference to the reward; is about