Vietnam Mapped: A Ridesharing Company's Open Data Initiative

Be Group is a fast-growing ridesharing company in Vietnam. Since joining Mapillary, they have contributed almost 9 million images and 27,000 km, greatly improving street-level imagery coverage in the country’s two biggest cities. In this post we explore how they've done this and what motivates them to collect street-level imagery.
Edoardo Neerhut
16 May 2024

Vietnam - An economy on the move

Be Group was established in 2018 as the developer of a “super app” that provides a range of daily life services, including ridesharing, food delivery, non-food delivery, telecommunications, and finance. The ridesharing (beBike, beCar, & beTaxi) and delivery (beFood & beDelivery) aspects of their business are dependent on maps, leading Be Group to quickly establish a dedicated mapping team. The BeMaps project was initiated in late 2022 by Ms. Vu Hoang Yen, Mr. Nguyen Tran Thanh - CEO & CTO of Be Group and Mr. Do Minh Phung - Maps Product Ops Lead at Be Group.

These maps are accessed by the user and the riders/drivers forming an essential part of the user experience. The maps team at Be Group uses a variety of methods to continually improve and update these maps, but it isn’t easy keeping map data current and accurate.

Vietnam has been one of the fastest growing economies in the world over the last decade (according to the IMF), averaging about 6% real GDP growth annually. Maps are hard to keep up to date at the best of times, but a fast-growing economy amplifies these challenges. Businesses open and close at a faster rate, city blocks are rezoned, buildings are knocked down and replaced, and already large cities such as Hanoi and Ho Chi Minh City continue to expand. In addition to rapid changes in Vietnam, familiar challenges such as urban density and narrow streets continue to complicate road network access for mappers. Be Maps’ solution is to utilize open source maps, their network of drivers, and street-level imagery to keep their maps consistently fresh.

Collecting imagery

For a number of key reasons, Be Group elected to use OpenStreetMap as its primary map provider. One major factor behind this is that they can update the map easily themselves, while also benefiting from contributions made by other individuals and organizations mapping Vietnam. Be Group started collecting street-level imagery in April 2023, using this wealth of information to translate the reality on the ground into useful map data.

For street-level image collection, their team opted to use the GoPro Max 360º camera, an affordable, easy to use option that can be mounted to the top of a rider’s helmet. Working with a team of 3 riders driving an average of 5-8 hours a day on scooters, Be Group was able to map almost the entirety of Hanoi and Ho Chi Minh City. The scooters allow riders to navigate many of the narrower laneways of Vietnamese cities that cars are unable to access. This meant they were able to collect map data that often gets ignored with traditional map data collection techniques.

Street-level imagery is just one part of their mapping process. As a ridesharing and delivery company, Be Group benefits from the extensive feedback generated daily from pickups and dropoffs. Some of this feedback comes directly from riders and drivers, and other feedback is more data driven with routes and pickup/drop-off points indicating the veracity of OpenStreetMap data — the company then contributes much of this data back to OpenStreetMap.

How they use imagery

Be Group collects street-level imagery for three main reasons.

1 ⛔️ Identify signs that restrict vehicular access.

Many roads in Ho Chi Minh City and Hanoi have entrance restrictions that sometimes apply only to cars. Keeping track of all these access restrictions is difficult, but can have a significant impact on trip times and driver/rider safety. The company uses Mapillary’s API to derive automatically detected traffic signs, which their team of mappers reviews and adds to OpenStreetMap. Since commencing their imagery collection, they have added 1,796 signs to OpenStreetMap.

2 🛣️ Determine & catalog the width of roads

While some roads in Vietnam explicitly prohibit cars, others have no formal indication but are nonetheless too narrow for a car to access. For the same reasons, Be Group uses driver feedback and imagery to check the reality on the ground and update the map accordingly.

3 🏪 Updating their database of pickup and drop-off points

Points of interest are one of the hardest challenges within mapping. They’re constantly changing and usually require ground truth to confirm. Mapillary is used to confirm place names and addresses, complementing the information Be Group gathers from thousands of trips. In addition to place names and addresses, pickup and drop-off points are another level of detail that determine the user experience. Large shopping malls, airports, hospitals, and similarly large points of interest can have multiple entrances and Be Group must determine which ones are most suitable for their users. Imagery can help to answer questions like “why are our customers starting their trips from a different address to the one we have for that point of interest?”.

Shared map data

OpenStreetMap has grown to be the incredible project it is because companies like Be Group can build services on top of it while contributing back with imagery and the insights they get from rides and deliveries. Feedback is an essential part of map making and ease of contribution is why more organizations are adopting OpenStreetMap and/or Overture, a multi-company collaboration to share and build products on top of open map data. Overture Maps builds off OpenStreetMap data, with the added benefit of data contributed by Overture members, including millions of places and 3D buildings. OpenStreetMap and Overture show the momentum accruing in the open mapping space and the benefits that result when a company like Be Group is able build a business on top of open data and contribute back their expertise of the Vietnamese market. We’re thrilled they’re contributing to Mapillary and allowing many others to map Vietnam in the process.