There is a "shady" in Didi Smart Dispatch. What will happen?

A famous recent example is that an artificial intelligence system named MogIA successfully predicted that Trump would become the president of the United States. This system was unveiled in 2004, and has successfully predicted three presidential elections before. Last year, the algorithm designed by the developer accurately predicted the last moment and rising momentum of the party again. Big data can also predict the weather, predict earthquakes, and even predict whether you will get sick. In terms of transportation, the ability of big data prediction is extremely important, which can predict when and where congestion will occur.

The key to big data prediction is enough high-quality data. At present, Didi's data volume in the transportation field ranks first in the world, with daily peak orders exceeding 20 million orders and daily processing data exceeding 2000TB, covering many dimensions such as traffic conditions, users' taxi information, drivers' driving behavior and vehicle data. Its huge real data can not only help predict road conditions, but also predict supply and demand. The more accurate the forecast of supply and demand, the more it can solve the problem of imbalance between supply and demand.

At present, the accuracy of supply and demand forecast after 15 minutes has reached 85%. Based on this high accuracy, the platform can dispatch drivers to meet the future taxi demand, effectively reducing the probability of imbalance between supply and demand in the future region. It is even conceivable that one day Didi will know exactly how many passengers there are in front of Li Gang Hotel and how much capacity there is nearby on Friday night, rainy or snowy days.

Route planning and ETA are two map technologies, which are the key to realize intelligent dispatching and will directly affect the experience of both drivers and passengers.

It is the core of dispatching order to predict the future road conditions and realize the path planning from point A to point B through massive historical data. Engineers carry out algorithms around the lowest price, the highest driver efficiency and the best traffic system operation efficiency.

ETA refers to estimating the travel time required for any starting point and ending point, which requires accuracy. Didi is the first company in China to successfully apply machine learning to ETA. Machine learning is the key technology to solve "efficient order matching" and "driver capacity scheduling". Now Didi ETA can predict the duration of each single trip and the waiting time before each intersection. With this technology, transportation capacity can be better scheduled at a more appropriate time.

Calling a car with Didi is not the same logic as searching. The information such as goods and information on the Internet stays there statically, and the calculation method is just to dig out the goods and information, while the calculation of Didi is similar to dynamic shooting. Vehicles are always moving, so it is necessary to give passengers an optimal choice among many moving vehicles, not only distance, but also time. That is, maximize platform efficiency and user experience. Intelligent dispatch predicts the number of orders and drivers, and then realizes the above-mentioned optimal matching through large-scale distributed calculation.