Didi Global. has been granted a patent for a method utilizing deep reinforcement learning for vehicle repositioning in ride-sharing platforms. The method involves analyzing vehicle location, evaluating potential paths, and selecting the optimal route based on expected rewards from a trained neural network. GlobalData’s report on Didi Global gives a 360-degree view of the company including its patenting strategy. Buy the report here.
Smarter leaders trust GlobalData
According to GlobalData’s company profile on Didi Global, V2I communication was a key innovation area identified from patents. Didi Global's grant share as of July 2024 was 39%. Grant share is based on the ratio of number of grants to total number of patents.
Vehicle repositioning using deep reinforcement learning
The patent US12061090B2 outlines a method and system for vehicle repositioning within a ride-sharing platform. The process begins by obtaining the current location of a vehicle and identifying a set of potential paths that originate from this location, ensuring that each path adheres to a specified maximum length. A trained deep value-network, which consists of two output branches—conditional and marginal value functions—is employed to calculate expected cumulative rewards for these paths. The conditional value function predicts values based on the vehicle's current state and a chosen action (either repositioning or remaining idle), while the marginal value function estimates expected values based on the current state alone. The network is trained using historical driver data, distinguishing between positive samples (repositioning actions) and negative samples (idle actions). The best path is then selected through a heuristic tree search based on these expected rewards, followed by delivering navigation instructions to the vehicle.
Additionally, the method incorporates a grid-world representation of the geographical area, where the vehicle's current location is mapped to a grid cell. This grid structure aids in excluding inaccessible paths and determining the next steps along the best path. The repositioning process can be triggered after a vehicle has been idle for a predetermined duration, and it may also consider the current time step to recommend future repositioning actions. The deep value-network further evaluates the repositioning cost and the future value of potential new locations based on dispatch probabilities and cumulative rewards. The system is designed to operate within an offline batch reinforcement learning framework, enhancing its efficiency in making repositioning decisions.
To know more about GlobalData’s detailed insights on Didi Global, buy the report here.
Data Insights
From
The gold standard of business intelligence.
Blending expert knowledge with cutting-edge technology, GlobalData’s unrivalled proprietary data will enable you to decode what’s happening in your market. You can make better informed decisions and gain a future-proof advantage over your competitors.
GlobalData, the leading provider of industry intelligence, provided the underlying data, research, and analysis used to produce this article.
GlobalData Patent Analytics tracks bibliographic data, legal events data, point in time patent ownerships, and backward and forward citations from global patenting offices. Textual analysis and official patent classifications are used to group patents into key thematic areas and link them to specific companies across the world’s largest industries.
