Uber Rival Didi Launches New Reinforcement Learning Algo To Improve Ride Hailing Efficiency

This reinforces the growing realization that Chinese tech companies are no longer simply copying Us or European enterprises, especially with regard to artificial intelligence and machine learning.

They are innovating in ways that surpass the performance of the first movers. JL

MIT Technology Review comments and Bernard Marr comments in Forbes:

The company’s current dispatching algorithm has two parts: a forecasting system that predicts how rider demand changes over time, and a matching system that assigns drivers to jobs on the basis of those predictions. It has served the company well, but can be inefficient. If the patterns of driver supply and rider demand change, the forecasting model needs to be retrained to continue making accurate predictions. Moving to an RL approach solves this by collapsing both parts into one: with every subsequent piece of data, the algorithm learns to dispatch drivers more efficiently.
Didi, China’s Uber equivalent, has been testing out a new algorithm for assigning drivers to riders in select cities.
The dispatching system uses reinforcement learning (RL), a subset of machine learning that relies on penalties and rewards to get “agents” to achieve a clear objective. In this case, the agents are the drivers and the rewards are their payments for completing a ride.
The company’s current dispatching algorithm has two parts: a forecasting system that predicts how rider demand changes over time, and a matching system that assigns drivers to jobs on the basis of those predictions.
It has served the company well thus far, but it can be inefficient. If the patterns of driver supply and rider demand change, the forecasting model needs to be retrained to continue making accurate predictions.
Moving to an RL approach solves this problem by collapsing both parts into one: with every subsequent piece of data, the algorithm learns to dispatch drivers more efficiently. That allows it to keep evolving with changing supply and demand, without any need to retrain. A/B tests between the old and new algorithms in a handful of cities have confirmed that the new one is indeed more efficient.
Didi is now planning a gradual roll-out of the new dispatching system to cities in China, though an exact time line hasn't been set. Tony Qin, the AI research lead for the company’s US division, told MIT Technology Review that the company may continue to conduct A/B tests between its different algorithms for different locations and use the one that produces the most efficient results.
The RL algorithm may not always be the best one, Qin said. It largely depends on the city’s supply and demand patterns. In the meantime, the company is also developing another RL dispatching algorithm, with different agents and rewards, to add to its arsenal.
Chinese company, Didi Chuxing may be known by most as the world’s largest ride-sharing company with a goal “to build a better journey,” but its vision reveals its future ambitions: “to become a global leader in the revolution in transportation and automotive technology.” With significant investment in artificial technology, Didi which means “beep beep” in Mandarin (like a car’s horn), is focused on staying ahead of the competition.
Harvard-educated Jean Liu, president of Didi Chuxing, is focused on growing the global footprint of the $56 billion-company that she leads. In China, the company has 550 million registered customers in more than 400 cities and delivers 30 million rides per day but its reach extends to Australia, Brazil, Japan and Mexico as well as Southeast Asia, India, Europe and Africa through various partnerships.
Didi employs 7,000 people, nearly half who are engineers and data scientists, and they continue to recruit other tech professionals to support its artificial intelligence labs, autonomous vehicles and other tech operations. On the Didi jobs page, Dr. Fengmin Gong, head of Didi Labs and Vice President of Didi Research Institute is quoted, “Our top priority is to continue to attract and retain the next generation of technology leaders from all backgrounds. We’re looking for brilliant, innovative minds to join us in solving some of the world’s toughest problems.”
The company currently has no rival in China; it battled Uber’s operations
in China and won. Uber entered China in 2013, and the two companies poured billions into fighting one another for the country’s ride-share business until pressure from Uber’s investors to stop losses caused the company to agree to a deal where Uber China’s operations were absorbed by Didi in return for Uber getting an 18 percent stake in Didi.
Innovation to Revolutionize the World’s Transportation and Automotive Technology
Didi was already a preferred tech investment for the likes of Alibaba, Tencent and Apple who collectively gave the company $13 billion as well as SoftBank, a Japanese tech conglomerate. In a new funding round that brought in an additional $4 billion, the company announced this new investment would be used for the company’s international expansion, artificial intelligence technology and green initiatives such as new energy vehicle networks.
In the quest to solve the world’s transportation challenges with smart traffic innovation, Didi has three artificial intelligence research centers, one in Silicon Valley and two in Beijing where research is done on natural language processing, computer vision and deep learning.
With the tremendous amount of data Didi collects every day, they are in a unique position to tackle traffic congestion and optimize navigation routes with AI technology. They launched the Didi Smart Transportation Brain that combines video camera and sensor data from Didi’s cars with data from the government and other partners. The objective is to create a city traffic management system powered by AI and cloud technology. The idea is that this will ultimately will result in smart traffic lights and monitoring systems that can be used in any metropolis with road congestion.
New technology to enhance the rider’s experience such as an app-based augmented reality (AR) navigation service that helps passengers find their way through buildings, malls and train stations to reach a vehicle pick-up location, illustrate Didi’s dedication to remaining the transportation leader. In cars, a voice-activated digital assistant offers a
wide range of services including audio and video content as wells as locations for fuel, recharging or repair services.
This same spirit of transportation innovation drives Didi’s efforts in autonomous vehicles, electric vehicles and redesigning vehicles for passengers who don’t need to drive. To ease the customer experience for passengers who use their apps in multiple countries, the company created a “roaming passport” when it launched in Japan that allowed them to use one app in any country they were in.
The company did experience a few bumps this past year from lackluster financial reports to a shift to prioritize safety over growth that slowed down their ambitious goals. In response to safety concerns following the murder of a Didi passenger and the rape and murder of a second rider this year, Didi announced new security features. These include random biometric ID testing for drivers and SOS buttons within the Didi driver and passenger app that go directly to police rather than a Didi call center.
And, since Uber China operations were absorbed in 2016, Didi hasn’t faced a competitor. That will soon change if Meituan-Dianping, China’s on-demand services provider, decides to move forward with its intentions to expand to the ride-sharing sector.
Didi Chuxing experienced meteoric influence since it was founded in 2012. If its leadership gets what it wants then the company could become as ubiquitous in the West as Amazon, Google and Facebook are today.