OVER the past few years, the media has focused a lot on how companies the likes of Uber, Grab and Airbnb have exploited the so-called ‘sharing economy,’ highlighting the fact that they don’t own assets but are still able to dominate the transportation and accommodation industries respectively.
But as these companies hog the headlines insofar as their valuations and growth rates are concerned, unbeknownst to many, they are also highly-tuned, data intensive-driven companies.
Being data driven means that they are able to pour resources into helping to solve one of society’s most challenging problems – that of traffic congestions in the countries they operate in.
Last April, Southeast Asian-based (SEA) Grab launched the OpenTraffic initiative in Malaysia, an effort aimed at providing traffic data from Grab’s GPS data streams to address traffic congestion and improve road safety in major Malaysian cities. The initiative was done in collaboration with the Malaysia Digital Economy Corporation Sdn Bhd (MDEC) and the World Bank Group.
Essentially, OpenTraffic provides Malaysia’s traffic management agencies and city planners access to an open dataset to better manage traffic flow and make investment decisions on local transport infrastructure.
The initiative is provided at no cost to governments via an open data licence. In practice, OpenTraffic translates Grab drivers’ GPS data into anonymised traffic data, to map traffic speeds on roads for analysing traffic congestion peak patterns and travel times.
The platform is designed to assist traffic management agencies in easing traffic flow, particularly within dense urban areas. Local government agencies can use the data to enhance existing traffic management systems such as optimising traffic light control and coordination.
Head of engineering for Grab Ditesh Gathani said the Open Traffic initiative started about two-and-a-half years ago in the Philippines.
At that time, the local authorities in the Philippines were facing a quandary as to how to manage their growing traffic woes, and it turned to the World Bank to help them solve these challenges, Ditesh said.
According to the World Bank, congestion in metropolitan Manila costs the economy more than US$60 million (RM237.14 million) per day, and it is not atypical to spend more than two hours to travel 8km during the evening commute there.
Conventional methods of collecting traffic data were either too capital intensive or too slow and inaccurate. This is when the World Bank turned to ride-hailing companies such as Grab and others such as Easy Taxi and Le.Taxi.
Combined together, these three ride-sharing companies cover more than 30 countries and millions of customers and are working with the World Bank and its partners to make traffic data derived from their drivers’ GPS streams available to the public through an open data licence.
“They [the Philippines] had traffic congestion and just needed help, and we [Grab] gave World Bank and its partners the data, and we can work together to visualise where the congestion is at,” he said.
Ditesh claimed the project was really successful in Cebu, Philippines and after that, the World Bank started expanding to other cities and they expanded into Indonesia and then to Malaysia.
“So, we are working with the World Bank on this, and supporting their programme. I believe they have plans to expand it to the rest of SEA and possibly out of the region too.”
Tech drives Open Traffic
Underpinning the Open Traffic initiative is Grab’s usage of machine learning models to help chart out traffic congestions patterns in real time, in order that users of the data may build visualisation models and understand traffic patterns.
Machine learning is a subset of artificial intelligence (AI), where software is taught to recognise patterns through the observation of massive data sets automatically, often without being explicitly programmed by humans.
Its goal is to automatically allow computers to take a certain action or draw certain conclusions from a large set of data, where the decisions made are automatically refined and evolved given more data analysed.
Ditesh said the technology powering the Open Traffic initiative is the same technology that Grab uses internally to optimise the supply and demand of drivers being called by users of the Grab app.
“The sharing of GPS data, all anonymised and aggregated, is aimed at giving government and city planning officials better data so that they can visualise where the congestion is, and plan traffic flow accordingly.
Ditesh explained that this is in line with Grab’s long-term goal to reshape cities in order that they can be lot safer and more efficient. He also revealed that within Grab, it has a team of data scientists, whose job is to just figure out these challenges.
Asked how difficult this process is, Ditesh said the problems are hard technical problems to solve. The process involves analysing not just the data trove that it has for today, but over the past few years, which runs into petabytes.
“When we looking at how to optimise traffic, in cities, there are general patterns as to how people go about,” he explained. “It took up to a year to build the underlying infrastructure to mine the data, analyse the traffic patterns, to power this kind of real-time visualisation.
“Coding is the easiest part of the equation, because by the time you need to code, you already have an idea of what you’re doing.
“The hardest part is what the code wants to do and that involves a lot of analysis and it’s open ended because you don’t know what kind of patterns you’re looking for in the data, and this takes a lot of intuition,” he said.
Ditesh also revealed that much of the machine learning expertise Grab has acquired is built in house because of the scale at which it operates.
“Also, no solution offered by any machine learning vendor is a perfect fit for our needs. We may use certain vendors’ products and start off with the off-the-shelf solution but in the end, we need to fine tune the solution to meet our needs.”
Quizzed as to what Grab’s plans are going forward insofar as machine learning is concerned, Ditesh said machine learning tools and models can enable a whole class to benefit the users and that it will continue to pursue this objective, but he didn’t reveal any details about Grab’s plans for the future.
Ditesh however said at present, Grab uses machine learning in combination with telematics to track driver behaviour. For example, the system can be used to track whether a driver is swerving due to dangerous driving behaviour or if a driver is sleepy.
Grab has built data science models that can show this data and alert drivers, via SMS, to warn them that they are not driving safely, Ditesh said. Using this, Ditesh claims that Grab drivers are six times safer compared to the national average.
“So, machine learning can be used in a wide variety of things, and safety is one good use case for machine learning within Grab.”
As for the future of the Open Traffic project, Ditesh said that currently, all Grab has done is to provide the anonymised and aggregated data to MDEC, and this year, the company expects to advance on the project.
“Grab is merely providing the data to MDEC in a bid to show them the extent of what it can do.
“The Philippines was a good test case because we had some good examples of how it can be used for road mapping and road safety,” Ditesh explained. “We haven’t gone to a second phase yet but we’re in discussions to do so to take it further with MDEC and other transportation-related authorities such as SPAD (Land Public Transport Commission).”
- January 2019(48)
- December 2018(46)
- November 2018(48)
- October 2018(76)
- September 2018(55)
- August 2018(63)
- July 2018(74)
- June 2018(64)
- May 2018(65)
- April 2018(76)
- March 2018(82)
- February 2018(65)
- January 2018(80)
- December 2017(71)
- November 2017(72)
- October 2017(75)
- September 2017(65)
- August 2017(97)
- July 2017(111)
- June 2017(87)
- May 2017(105)
- April 2017(113)
- March 2017(108)
- February 2017(112)
- January 2017(109)
- December 2016(110)
- November 2016(121)
- October 2016(111)
- September 2016(123)
- August 2016(169)
- July 2016(142)
- June 2016(152)
- May 2016(118)
- April 2016(60)
- March 2016(86)
- February 2016(154)
- January 2016(3)
- December 2015(150)