Abstract: Systems and methods for training artificial intelligence models based on sequences of image data are disclosed. The techniques described herein include generating, using an artificial intelligence model, a respective classification and a respective bounding box for an object depicted in each image of a sequence of images captured during operation of an autonomous vehicle; tracking the object in the sequence of images based on the respective bounding box of each image of the sequence of images and a tracking identifier corresponding to the object; determining a correction to the respective classification of an image of the sequence of images responsive to tracking the object in the sequence of images; and training the artificial intelligence model based on the correction.

Abstract: A method of controlling an autonomous vehicle, including: collecting perception data representing a perceived environment of the vehicle using a perception system on board the autonomous vehicle; comparing the perception data collected with digital map data; and modifying operation of the vehicle based on an amount of difference between the perception data and the digital map data.

Abstract: A method of controlling an autonomous vehicle, including: collecting perception data representing a perceived environment of the vehicle using a perception system on board the autonomous vehicle; comparing the perception data collected with digital map data; and modifying operation of the vehicle based on an amount of difference between the perception data and the digital map data.

Abstract: Systems and methods for path planning for autonomous vehicles err on the side of a safer road position and safer type of collision if one were to occur. An autonomous vehicle perception module detects target objects within a region of interest (ROI). An autonomous vehicle processor estimates relative velocity between the ego vehicle and target objects, and estimates mass of target objects. Collision-aware path planning calculates a cost function that assigns higher costs for paths that take the vehicle close to objects that have a large velocity difference and higher costs for objects with large estimated mass. Path planning provides a cost map that yields a path that has appropriate buffer distances between the autonomous vehicle and surrounding objects. In the event the ego vehicle balances buffer distance margins to multiple surrounding target objects, target objects that would create more severe collisions are given greater buffer distance.

Abstract: A method may include monitoring, by a processor associated with an autonomous vehicle, a status of a traffic light. The method may include monitoring, by the processor, a movement status of at least one vehicle within a same lane as the autonomous vehicle, the at least one vehicle having a stationary position in relation to the traffic light. The method may include executing, by the processor, a computer model to predict a change of movement status for the at least one vehicle in accordance with the status of the traffic light. The method may include in response to the computer model predicting a change of movement status for the at least one vehicle from the stationary position to a moving status, instructing, by the processor, the autonomous vehicle to increase a torque value of the autonomous vehicle at a predetermined time before the change of the movement status.

Abstract: Systems and methods for training and executing machine learning models to generate lane index values are disclosed. A method includes identifying a set of image data captured by at least one autonomous vehicle when the at least autonomous vehicle is positioned in a lane of a roadway and respective ground truth localization data; determining a plurality of lane index values for the set of image data based on the ground truth localization data; labeling the set of image data with the plurality of lane index values, the lane index values representing a number of lanes from a leftmost or rightmost lane to the lane in which the at least one autonomous vehicle was positioned; and training, using the labeled set of image data, a plurality of machine learning models that generate a left lane index value and a right lane index value as output.

Abstract: A method of detecting an object in a path of an vehicle using a LiDAR system includes emitting a LiDAR signal with the LiDAR system; receiving the LiDAR signal with the LiDAR system; determining a glancing angle distance; determining the LiDAR signal is received from beyond the glancing angle distance based on receipt of the LiDAR signal; and classifying the LiDAR signal as a return from an object based at least in part on the LiDAR signal coming from a distance beyond the glancing angle distance.

Abstract: A method comprises identifying a set of image data captured by at least one autonomous vehicle when the at least one autonomous vehicle was positioned in a lane of a roadway, and respective ground truth localization data of the at least one autonomous vehicle; determining a total number of lanes for the roadway; labeling the set of image data with the total number of lanes for the roadway; and training using the labeled set of image data, a machine learning model, such that the machine learning model is configured to predict a new total number of lanes for a new roadway as output.

Abstract: A method, comprises identifying a set of image data captured by at least one autonomous vehicle when the at least one autonomous vehicle was positioned in a lane of a roadway, and respective ground truth localization data of the at least one autonomous vehicle; determining a plurality of lane width values for the set of image data; labeling the set of image data with the plurality of lane width values, the plurality of lane width values representing a width of a lane in which the at least one autonomous vehicle was positioned; and training using the labeled set of image data, a machine learning model, such that the machine learning model is configured to predict a new lane width value for a new lane as output.

Abstract: Systems and methods for training and executing machine learning models to generate lane index values are disclosed. A method includes identifying a set of image data captured by at least one autonomous vehicle when the at least autonomous vehicle is positioned in a lane of a roadway and respective ground truth localization data; determining a plurality of lane index values for the set of image data based on the ground truth localization data; labeling the set of image data with the plurality of lane index values, the lane index values representing a number of lanes from a leftmost or rightmost lane to the lane in which the at least one autonomous vehicle was positioned; and training, using the labeled set of image data, a plurality of machine learning models that generate a left lane index value and a right lane index value as output.

Abstract: Disclosed herein are systems and methods for operating a vehicle. In an embodiment, a system can identify a maximum distance bound based on one or more objects around a vehicle; for each of a plurality of candidate trajectories for the vehicle, determine a velocity from the maximum distance bound at an ending time of the candidate trajectory; determine an available distance for the candidate trajectory as a function of the determined velocity at the ending time of the candidate trajectory and a comfort deceleration parameter; determine a target velocity for the candidate trajectory; and determine a velocity difference between the target velocity and a final velocity of the candidate trajectory at the ending time of the candidate trajectory; select a first candidate trajectory based on the velocity difference; and operate the vehicle based on the selected first candidate trajectory.

Abstract: Systems and methods for training and executing machine learning models to generate lane index values are disclosed. A method includes identifying a set of image data captured by at least one autonomous vehicle when the at least autonomous vehicle is positioned in a lane of a roadway and respective ground truth localization data; determining a plurality of lane index values for the set of image data based on the ground truth localization data; labeling the set of image data with the plurality of lane index values, the lane index values representing a number of lanes from a leftmost or rightmost lane to the lane in which the at least one autonomous vehicle was positioned; and training, using the labeled set of image data, a plurality of machine learning models that generate a left lane index value and a right lane index value as output.