IBM Nairobi Lab’s First Offering is a Traffic-Dodging Mobile App

Nichole Sobecki/Feature Photo/IBM

Debates about how best to avoid Nairobi traffic can take nearly as long as a drive across town. The city has three dozen traffic cameras downtown, but that’s not enough information for a city of over three million people. Traffic costs the city US $600 000 a day, by one estimate. IBM’s Nairobi lab, in beta since a year ago, tackled traffic early on and today launched a mobile application to help drivers avoid traffic.

The app, called Twende Twende, meaning something like “Let’s go” in Swahili, bases its recommendations on a central system that uses image recognition algorithms to process the traffic camera feeds and a separate algorithm to predict traffic on streets not covered by the cameras. Users can get recommendations via SMS or on a map interface.

In answer to the obvious question of what advantage this has over Google Maps’ traffic feature, the lab’s chief scientist, Osamuyimen “Uyi” Stewart, explained that IBM considers observational data from cameras more reliable than Google’s crowd-sourced data. Time, and perhaps Nairobi-based Spectrum readers, will tell. The app is available through Safaricom and Airtel, two Kenyan mobile service providers.

Whatever the lab learns from this and other Nairobi-grown innovations should be useful elsewhere in Africa and perhaps globally, Stewart says. The app is one of the first results of the lab’s first year in operation, and the fact that the company is promoting it on its official opening today, together with another data analysis product, signals its emphasis on data analysis in the developing world.

Last month Stewart told journalists that IBM is trying to position its research to serve the growing number of African countries transitioning from resource extraction to service economies. “Data is the next generation of resources in Africa,” he declared.

The lab is also involved in setting up systems that make it easy to collect useful economic data, such as agricultural performance, weather, pricing, and finding ways to capitalize on the information. Farmers, for example, could obtain access to certain data themselves in return for participating in mobile phone surveys.

With huge growth in demand for middle-class services across much of Africa, the lab has its work cut out for it. If Twende Twende works as promised, at least IBM researchers will find it easier to get to and from that work.

First published at IEEE Spectrum’s Tech Talk blog: [html] [pdf]