HERE Global

An approach is provided for providing localization through multiple triangulation processes. The approach, for instance, involves detecting a plurality of objects from a set of images, performing 3-point triangulation on the plurality of objects to generate a 3-point triangulation result, and performing one or more triangulations on the 3-point triangulation result to generate one or more triangulation results. The one or more triangulations is 4-or-more-point triangulation, 2-point triangulation, or a combination thereof. The one or more triangulation results is a 4-or-more-point triangulation result, a 2-point triangulation result, or a combination thereof. The approach further involves using a clustering model to consolidate the 3-point triangulation result and the one or more triangulation results as a consolidated localization result and outputting the consolidated localization result.

Embodiments described herein relate to contextually recommend parking locations to improve the privacy of visited destinations. Methods may include: receiving a routing request from a user for a route to a destination Point-of-Interest, POI; determining a sensitivity of the destination POI; analyzing parking locations proximate the destination POI to determine a level of privacy associated with the parking locations POI; selecting at least one parking location based on the level of privacy associated with the at least one parking location; generating at least one candidate route to the at least one parking location; and calculating a privacy rating associated with the at least one candidate route and the at least one parking location based on a likelihood of associating the at least one candidate route and the at least one parking location with the destination POI.

Embodiments described herein relate to anonymizing of trajectories based on an anticipated probe data density at a location associated with the trajectory at a time corresponding to a time the trajectory was generated. Methods may include: identifying a plurality of geographic sub-areas of a geographical region; determining a plurality of timeslots associated with the geographical region using temporal partitions; identifying, for each timeslot of each geographic sub-area, a probe data density; determining anonymization parameters for an anonymization strategy based on the probe data density; receiving probe data as a trajectory from within a first geographic sub-area at a first timeslot; anonymizing the trajectory using the anonymization strategy with anonymization parameters associated with the probe data density corresponding to the first geographic sub-area at the first timeslot to form an anonymized trajectory; and providing the anonymized trajectory to a service provider for provision of location-based services.

An approach is provided for path-based map matching. The approach involves, for instance, executing a first pass map-matching process to determine on-road candidate matches for observation points of the trajectory. The approach also involves determining emission probabilities of the candidate matches and transition probabilities between on-road candidates of subsequent observations, and determining a most probable path based on the emission and transition probabilities. The approach further involves determining a set of a predetermined number of on-road candidate matches from the most probable path. The approach further involves marking at least one match in the set as not matched based on determining that an alternate transition with a higher transition probability than the at least one transition exists. The approach further involves redetermining the most probable path using a second pass map-matching process based on the marked match.

Example embodiments describe a computer-implemented method, an apparatus, and a computer program product for generating personalized content for users traversing a route. The method may obtain a route and its content attention profile, which assigns an estimated content attention level to each road link based on user attention levels corresponding to various road link features. A control prompt may be generated based on this profile and a first content data file containing text, audio, or video data. The control prompt may instruct adjusting the detail level of the content. A generative artificial intelligence component may process the first content data file and control prompt to generate a second content data file. The second content data file may be presented on a user device. This method may provide an adaptive entertainment experience for users along their route, potentially enhancing engagement and enjoyment during travel.

Methods, apparatuses and computer program products for updating map data are disclosed. The method for updating a map comprises obtaining first map data to be updated, obtaining second map data to be used for updating the first map data, determining a matched second map road segment of a road of the second map data matched to a corresponding matched first map road segment of a road of the first map data, registering the matched second map road segment to the matched first map road segment; and updating the first map data by updating the first map road segment.

A road sign detection method comprises obtaining an unlabeled image of a road segment and executing a neural network trained to identify road signs within the unlabeled image. The neural network comprises a first model configured to localize an object in the unlabeled image. The neural network may further comprise a second model configured to generate classification data associated with the cropped image, based on content of the cropped image, where the classification data corresponds to one or more appearance-based attributes of the cropped image and also to determine a category of the cropped image, based on the classification data of the cropped image. The neural network may comprise a third model configured to generate road sign label data based on the determined category of the cropped image, where the road sign label data represents the identified road sign. The method further comprises outputting the road sign label data.

An approach is provided for matching to directed graph representation of lanes of a road topology network of a digital map. The approach, for instance, involves determining a plurality of lane group centerlines respectively representing a plurality of lane groups of a road network in a geographic database. The approach also involves for each lane group centerline of the plurality of lane group centerlines, determining one or more edges having a pair of node identifiers. The approach further involves inserting the one or more edges and the pair of node identifiers into a lane group routing graph, wherein the lane group routing graph is a directed graph. The approach further involves matching an input directed graph to the lane group routing graph, wherein the input directed graph represents a set of trajectories, and providing a map matching result based on the matching.

An approach is provided for lane-level direction-of-travel detection from overhead imagery. The approach, for instance, involves processing overhead imagery data of a road network to determine a plurality of features indicative of a direction of travel of at least one road lane of the road network. Each feature of the plurality of features indicates a candidate direction of travel. The approach also involves determining the direction of travel of the at least one road lane based on a weighted voting mechanism applied based on the candidate direction of travel of each feature. The weighted voting mechanism weighs the candidate direction of travel of each feature based on a feature type of each feature. The approach further involves associating the determined direction of travel as an attribute of the at least one road lane in a geographic database.

An approach is provided for map matching a vehicle path network to a road topology network of a digital map. The approach, for instance, involves aggregating vehicle trajectory data into a vehicle path network that is a directed graph comprising a plurality of nodes. The approach also involves segmenting the vehicle path network into a plurality of linear features. Each linear feature of the plurality of linear features connects two nodes of the plurality of nodes in the vehicle path network. The approach further involves using a path-based map matcher to separately match each linear feature by incorporating context of the vehicle path network to a map topology of a digital map to generate map matching results. The approach further involves aggregating the map matching results of each linear feature into an overall map matching result for the directed graph, and providing the overall map matching result as an output.