Abstract: A system comprises a processor and a memory storing instructions. The processor receives an image for processing using a reinforcement learning based agent comprising a neural network trained using a reward function. The image includes N lane lines of a roadway, where N is a positive integer. The instructions configure the processor to traverse the image using the agent at least N times from a first end of the image to a second end of the image by: incrementally moving the agent from a first side of the image to a second side of the image after each traversal; and maximizing rewards for the agent using the reward function during each traversal of the image using the agent. The instructions configure the processor to identify the N lane lines of the roadway as a single cluster of lane lines after traversing the image using the agent at least N times.

Abstract: System and methods for estimating a centerline of a road that separates traffic moving in opposite directions include aggregating a data set from each of a plurality of vehicles traversing the road over a period of time as telemetry data. Each data set of the telemetry data indicates a location and a heading. The method includes clustering the data sets of the telemetry data based on the heading indicated by each data set, and identifying a separator to separate the data sets indicating a first heading from the data sets indicating a second heading, opposite to the first heading. The centerline is estimated based on applying a spatial smoothing to the separator.

Abstract: A system and a method of generating a map for navigating a vehicle. The system includes a remote processor. The remote processor is configured to determine an anchor point for a location of a road segment based on data from at least one data source, place the anchor point within an aerial image of the road segment, predict a boundary marking of the road segment on the aerial image based on the anchor point, and provide the boundary marking to the vehicle. A vehicle processor navigates the vehicle along the road segment using the boundary marking.

Abstract: A method for updating a map including receiving a first image depicting a geographical area including a first roadway and an occluded area, determining a location of the first roadway segment in response to the first image, receiving a plurality of vehicle telemetry data associated with the first roadway segment and a second roadway segment within the occluded area, updating a map data with the location of the first roadway, determining a location of the occluded area in response to the first image and the plurality of vehicle telemetry data associated with the second roadway segment, requesting an alternate data in response to determination of the location of the occluded area, determining a location of a second roadway segment in response to the alternate data wherein the second roadway segment was occluded in the first image, and updating the map data with the location of the second roadway segment.

Abstract: A system for inferring lane boundaries via vehicle telemetry for a road section is provided. The system includes a computerized remote server device operable to receive through a communications network sensor data describing lane markings upon roads bordering the road section, determine established lanes upon roads bordering the road section based upon the sensor data, receive through the communications network the vehicle telemetry generated by a plurality of vehicles traversing the road section, generate inferred lanes for the road section based upon the vehicle telemetry, match the inferred lanes to the established lanes to generate unified lane geometries, and publish the unified lane geometries.

Abstract: A system comprises a processor and a memory storing instructions. The processor receives an image for processing using a reinforcement learning based agent comprising a neural network trained using a reward function. The image includes N lane lines of a roadway, where N is a positive integer. The instructions configure the processor to traverse the image using the agent at least N times from a first end of the image to a second end of the image by: incrementally moving the agent from a first side of the image to a second side of the image after each traversal; and maximizing rewards for the agent using the reward function during each traversal of the image using the agent. The instructions configure the processor to identify the N lane lines of the roadway as a single cluster of lane lines after traversing the image using the agent at least N times.

Abstract: A method to detect a roadway edge includes calculating a first likelihood of a roadway edge from an aerial image of a roadway by shifting a centerline of the roadway perpendicular to the centerline and overlapping the centerline with image gradients. A second likelihood of the roadway edge is determined using a vehicle telemetry fitting a probability distribution to telemetry points along the roadway. The first likelihood of the roadway edge and the second likelihood of the roadway edge are fused to identify a final likelihood of the roadway edge.

Abstract: Systems and methods are provided for generating a map for use in controlling a vehicle. In one embodiment, a method includes: receiving, by a processor, aerial image data depicting an environment; processing, by the processor, the aerial image data with a plurality of trained deep learning models to produce a predicted radar map; and controlling the vehicle based on the predicted radar map.

Abstract: A method to detect a roadway edge includes calculating a first likelihood of a roadway edge from an aerial image of a roadway by shifting a centerline of the roadway perpendicular to the centerline and overlapping the centerline with image gradients. A second likelihood of the roadway edge is determined using a vehicle telemetry fitting a probability distribution to telemetry points along the roadway. The first likelihood of the roadway edge and the second likelihood of the roadway edge are fused to identify a final likelihood of the roadway edge.

Abstract: According to exemplary methods, automated image processing of images from order locations of a multi-lane drive-thru system is continually performed using an automated ordering system. It is automatically determined if a vehicle is present at any of the order locations based on the automated image processing. It is automatically determined if an order has been initiated into the automated ordering system. When the order has been initiated while the vehicle is present at any of the order locations, an automated sequencing processing is automatically begun. The automated sequencing processing determines an order pick-up sequence for picking up orders at pickup windows of the multi-lane drive-thru system. When the order has been initiated while the vehicle is not present at any of the order locations, a potential out-of-sequence warning is automatically associated with the order and output from the automated ordering system.

Abstract: Systems and methods to generate an augmented map used for autonomous driving of a vehicle involve obtaining images at a first point of view, and training a first neural network to identify and classify features related to an attribute in the images at the first point of view. A method also includes projecting the features onto images obtained at a second point of view, and training a second neural network to identify the attribute in the images at the second point of view based on the features. The augmented map is generated by adding the attribute to a map image at the second point of view.

Abstract: According to exemplary methods of training a convolutional neural network, input images are received into a computerized device having an image processor. The image processor evaluates the input images using first convolutional layers. The number of first convolutional layers is based on a first size for the input images. Each layer of the first convolutional layers receives layer input signals comprising features of the input images and generates layer output signals that include signals from the input images and ones of the layer output signals from previous layers within the first convolutional layers. Responsive to an input image being a second size larger than the first size, additional convolutional layers are added to the convolutional neural network. The number of additional convolutional layers is based on the second size in relation to the first size. The additional convolutional layers are initialized using weights from the first convolutional layers.

Abstract: A system for inferring lane boundaries via vehicle telemetry for a road section is provided. The system includes a computerized remote server device operable to receive through a communications network sensor data describing lane markings upon roads bordering the road section, determine established lanes upon roads bordering the road section based upon the sensor data, receive through the communications network the vehicle telemetry generated by a plurality of vehicles traversing the road section, generate inferred lanes for the road section based upon the vehicle telemetry, match the inferred lanes to the established lanes to generate unified lane geometries, and publish the unified lane geometries.

Abstract: System and methods for estimating a centerline of a road that separates traffic moving in opposite directions include aggregating a data set from each of a plurality of vehicles traversing the road over a period of time as telemetry data. Each data set of the telemetry data indicates a location and a heading. The method includes clustering the data sets of the telemetry data based on the heading indicated by each data set, and identifying a separator to separate the data sets indicating a first heading from the data sets indicating a second heading, opposite to the first heading. The centerline is estimated based on applying a spatial smoothing to the separator.

Abstract: In one example implementation according to aspects of the present disclosure, a computer-implemented method for generating map features includes receiving, by a processing device, aerial image data. The method further includes receiving, by the processing device, telemetry data. The method further includes performing, by the processing device, data fusion on the aerial image data and the telemetry data to generate map features. The method further includes controlling, by the processing system, a vehicle based at least in part on the map features.

Abstract: Systems and methods are provided for generating a map for use in controlling a vehicle. In one embodiment, a method includes: receiving, by a processor, aerial image data depicting an environment; processing, by the processor, the aerial image data with a plurality of trained deep learning models to produce a predicted radar map; and controlling the vehicle based on the predicted radar map.

Abstract: A smoke detection system for an interior of a vehicle is provided. The smoke detection system includes a detection module and a reporting module. The detection module, while the vehicle is being used as part of a car sharing or ride sharing service: receives a sensor signal from a smoke detector; compares data included in the sensor signal to smoke characteristic or smoke pattern data; based on the comparison, determines whether an occupant of the vehicle is smoking; and generates an alert signal if the occupant is smoking. The reporting module at least one of: based on the alert signal, visually or audibly provides a warning to the occupant to stop smoking; transmits the alert signal to a mobile device of the occupant to indicate that a smoking event has been detected; or transmits the alert signal to a network device of a service provider to report the smoking event.

Abstract: A system for partitioning a two-dimensional (2D) array and recognizing elements of the 2D array selects a location in an array comprising pixel data of the 2D array; and selects a direction in which to traverse the array from the selected location to segment the elements of the 2D array. The system uses a model and a reward/penalty function to guide the traversal through the array. The system changes direction when the data encountered while traversing the array in the selected direction represents one of the elements of the 2D array and continues in the selected direction when the data represents a joint between two elements of the 2D array. The system recognizes one or more of the elements of the 2D array segmented by traversing the array.

Abstract: Systems and methods of mapping traffic related objects. In one embodiment, a method includes: receiving image sensor data relating to an environment associated with a vehicle; receiving lidar sensor data relating to the environment associated with the vehicle; selecting, by a processor, at least one image region including a traffic related object from the image sensor data; selectively filtering, by a processor, the at least one image region based on the lidar sensor data; and generating, by a processor, a composite map including a representation of the traffic related object superimposed upon a defined map of the environment based on the selectively filtering.

Abstract: A system for partitioning a two-dimensional (2D) array and recognizing elements of the 2D array selects a location in an array comprising pixel data of the 2D array; and selects a direction in which to traverse the array from the selected location to segment the elements of the 2D array. The system uses a model and a reward/penalty function to guide the traversal through the array. The system changes direction when the data encountered while traversing the array in the selected direction represents one of the elements of the 2D array and continues in the selected direction when the data represents a joint between two elements of the 2D array. The system recognizes one or more of the elements of the 2D array segmented by traversing the array.