Abstract: A method of determining a parking route for a vehicle travelling on a road network within a geographic area is disclosed. A sub-network is determined comprising a subset of segments of an electronic map that are representative of roads within a predetermined walking time or distance of a destination location. Data indicative of a walking time or distance from the segment to the destination location is associated, at least with the segments of the sub-network representing roads having at least one associated parking space. A search algorithm having an associated cost function is used to explore the segments of the sub-network from an origin location to identify a plurality of candidate parking routes, wherein the cost for a given parking route is based on the probability of the vehicle successfully finding a parking space on the parking route and the expected cumulative travel and walking time or distance to the destination location should a parking space be found along the parking route.

Abstract: A method for generating guidance information is disclosed, which uses one or more machine learning algorithms to continually adapt and update a version of a model used to generate guidance instructions based on feedback data received from one or more users, such that the instructions generated by the model better fit the expectations or behaviour of the user or users. The method can be performed on a mobile navigation device, such that the model is adjusted to suit the needs of a single user. Alternatively, the method can be performed on a server based on feedback data from a plurality of users, and the updated model, once generated, transmitted to a plurality of mobile navigation devices to replace the existing model.

Abstract: A method of determining a parking route for a vehicle travelling on a road network within a geographic area is disclosed. A sub-network is determined comprising a subset of segments of an electronic map that are representative of roads within a predetermined walking time or distance of a destination location. Data indicative of a walking time or distance from the segment to the destination location is associated, at least with the segments of the sub-network representing roads having at least one associated parking space. A search algorithm having an associated cost function is used to explore the segments of the sub-network from an origin location to identify a plurality of candidate parking routes, wherein the cost for a given parking route is based on the probability of the vehicle successfully finding a parking space on the parking route and the expected cumulative travel and walking time or distance to the destination location should a parking space be found along the parking route.

Abstract: A method for generating guidance information is disclosed, which uses one or more machine learning algorithms to continually adapt and update a version of a model used to generate guidance instructions based on feedback data received from one or more users, such that the instructions generated by the model better fit the expectations or behaviour of the user or users. The method can be performed on a mobile navigation device, such that the model is adjusted to suit the needs of a single user. Alternatively, the method can be performed on a server based on feedback data from a plurality of users, and the updated model, once generated, transmitted to a plurality of mobile navigation devices to replace the existing model.