BEHAVIOR-CONTROLLED ROUTE PLANNING IN AUTONOMOUS MACHINE APPLICATIONS

摘要

In various examples, a machine learning model - such as a deep neural network (DNN) - can be trained to use image data and / or other sensor data as inputs to two- or three-dimensional trajectory points in world space, vehicle orientation and / or generate a vehicle condition. For example, sensor data representing an orientation, steering information and / or a speed of a vehicle can be collected and used to automatically generate a trajectory for use as basic truth data for training the DNN. If applied, the trajectory points, vehicle orientation, and / or vehicle condition can be used by a control component (e.g., a vehicle controller) to control the vehicle through a physical environment. For example, the control component can use these outputs of the DNN to determine a vehicle-specific control profile (e.g. steering, decelerating and / or accelerating) for controlling the vehicle through the physical environment.