System and process for end to end prediction of lane detection uncertainty

A system for end to end prediction of lane detection uncertainty within a host vehicle, comprising:

• the host vehicle, including:
• a sensor device for a host vehicle operable upon a road surface within a driving environment including lane markings, wherein the sensor device is operational to generate data related to the road surface within the driving environment, wherein the sensor device includes one of a camera device or a LiDAR device;
• a navigation controller including a computerized processor operable to:
• monitor the data including an input image from the sensor device;
• utilize a convolutional neural network to analyze the input image and output a lane prediction and a lane uncertainty prediction, wherein the convolutional neural network includes:
• a trained tiles prediction deep network performing image recognition upon the input image;
• and
• an error prediction head network initially trained using a per point association and error calculation, the per point association and error calculation including:
• first, associating a selected ground truth lane to a selected set of data points related to a predicted lane, and
• second, associating at least one point of the selected ground truth lane to a corresponding data point from the selected set of data points related to the predicted lane;
• and
• generate a commanded navigation plot based upon the lane prediction and the lane uncertainty prediction;
• and
• a vehicle controller operable to control the host vehicle based upon the commanded navigation plot.

A process for end to end prediction of lane detection uncertainty, comprising:

• within a computerized processor:
• initially training a convolutional neural network using a per point association and error calculation, the per point association and error calculation including:
• first, associating a selected ground truth lane to a selected set of data points related to a predicted lane;
• and
• second, associating at least one point of the selected ground truth lane to a corresponding data point from the selected set of data points related to the predicted lane;
• subsequently utilizing the convolutional neural network for operational use in inference of lane geometries including:
• monitoring data including an input image from a sensor device of a host vehicle generating the data related to a road surface in front of the host vehicle including lane markings;
• utilizing the convolutional neural network to analyze the input image and output a lane prediction and a lane uncertainty prediction;
• and
• generating a commanded navigation plot based upon the lane prediction and the lane uncertainty prediction;
• and
• controlling the host vehicle based upon the commanded navigation plot.