patOS
← Projects
27 August 2026 · PatOS

The Map That Remembers Where You’ve Been — and Finds Somewhere New

How Rides turns personal e-bike history and open map data into inspectable routes that favour roads you have not ridden yet.

Projectsridesroute-planningopen-data

Most route planners begin with a destination. Patrick’s Rides project begins with a more personal question: where haven’t you ridden yet?

Rides turns years of Bosch eBike Connect history into a living map of explored and unexplored roads. The project snapshot extracted through 7 July 2026 represents more than 1,100 mapped rides, over 7,700 kilometres and roughly 1.43 million GPS points. The result is not another generic list of popular loops. It is a map shaped by one rider’s actual history.

That distinction changes the job of the software. A conventional planner can optimize for distance, speed or popularity. Rides can also value novelty: it can see the roads already travelled and favour roads that would add something new.

Personal history as useful infrastructure

The first half of Rides is deliberately simple. It plots past rides on an interactive map, with filters for year and month, summary statistics and a heatmap. The core map artifact is packaged as a single browser-openable file, so the saved history and route interface do not depend on a separate application installation.

The second half uses that history to propose the next ride. Its offline routing engine works from an OpenStreetMap-derived road graph covering five Dutch provinces. The graph contains about 5.3 million nodes, and the route calculation adjusts for novelty, scenery and traffic rather than treating every road as interchangeable.

That produces several kinds of outing:

Every route can be exported as GPX for use with Bosch eBike Connect. Rides can also create day packs with printable maps, addresses and short landmark notes. That makes the route useful beyond the screen: it can move into the navigation tool a rider already uses and be carried as a PDF day pack.

Bosch’s own documentation confirms that GPX files can be imported into its route-planning tools and saved routes. Rides builds around that supported hand-off instead of pretending to replace the bike’s navigation ecosystem.

Open data, visibly credited

Rides is possible because several public knowledge systems can be combined without hiding where the information came from.

OpenStreetMap supplies the underlying road and place data under the Open Database License, created and maintained by OpenStreetMap contributors. CARTO provides map styling used in the project. OpenRailwayMap adds the rail context that makes station-linked routes legible. Nominatim resolves place names and addresses, while Wikipedia and Wikidata contribute short descriptions and verification clues for landmarks and stations.

The live Destination mode uses the public cycling instance at routing.openstreetmap.de, which is a community service rather than a PatOS-owned routing backend. That provenance matters: a useful interface should not erase the public infrastructure beneath it.

It also matters because sources imply constraints. OpenStreetMap is community-maintained, public routing services can change, and an attractive map tile is not the same thing as a guaranteed route. Rides exposes a practical composition of those sources; it does not turn them into certainty.

Constraints are part of the product

The pinned Rides source records a substantial route-quality pass: generated loops were checked for closure, target-distance tolerance and excessive overlap; later variants added maximum-distance caps, and 74 short JKT routes were held to the 35-kilometre limit. Those are source-verified project checks, not a claim that every future road condition has been validated in the field.

The project is similarly candid about its current boundaries. The offline graph covers five provinces rather than the whole world. It does not currently enforce one-way restrictions. Some large stations need careful snapping to the road graph. Live Destination routes have a weaker novelty signal than the precomputed modes and depend on external geocoding and routing services.

Those limitations are not footnotes to conceal. They describe where judgement remains with the rider.

Grace’s PatOS research on change-aware plans reached a related conclusion from travel research: useful assistance should preserve sources, constraints and fallbacks instead of presenting a brittle itinerary as finished truth. Applying that insight to Rides is a design hypothesis, not evidence of market demand. But it gives the project a sound direction: help someone explore more widely while keeping the route inspectable, exportable and theirs to accept.

A route planner with a memory

Rides shows what can happen when personal data is treated as material for a better tool rather than as a product to be harvested. Patrick’s own ride history supplies the memory. Open data supplies the navigable world. The software joins them to make the next outing less repetitive and more personal.

The most interesting feature is not any single route mode. It is the shift in perspective: the map no longer asks only where a road leads. It remembers whether you have already been there.

Sources and credits