A system and method for visual inertial navigation for augmented reality are described. In some embodiments, at least one camera of a wearable device generates a plurality of video frames. At least one inertial measurement unit (IMU) sensors of the wearable device generates IMU data. Features in the plurality of video frames for each camera are tracked. The plurality of video frames for each camera are synchronized and aligned based on the IMU data. A dynamic state of the wearable device is computed based on the synchronized plurality of video frames with the IMU data for each camera. Augmented reality content is generated and positioned in a display of the wearable device based on the dynamic state of the wearable device.