Grad student earns Outstanding Paper Award at RTSS 2021
The paper by Ding and his collaborators was recognized for its high-quality research contributions to RTSS, the premier conference in the field of real-time systems.
The award-winning work, Concurrent Order Dispatch for Instant Delivery with Time-Constrained Actor-Critic Reinforcement Learning proposes a time-constrained actor-critic reinforcement learning-based concurrent dispatch system called TCAC-Dispatch to enhance the long-term overall revenue and reduce the overdue rate.
The researchers designed a deep matching network with a variable action space, which integrates the state embedding (including route behaviors encoding) and actions embedding features into a long-term matching value. Then the actor-critic model tackles the concurrent order dispatch problem considering strict time constraints and stochastic demand-supply in instant delivery. An estimated time-based action pruning module is designed to ensure time constraints guarantee and accelerate the training as well as dispatching processes.
The team evaluated the TCAC-Dispatch with one-month data involved with 36.48 million orders and 42,000 couriers collected from one of the largest instant delivery companies in China. Empirical experiments were conducted on a data-driven emulator deployed on the development environment and results showed that the method achieved 22% of the increase in total revenue and reduced the overdue rate by 21.6%.
Ding contributed to the research in a number of ways. In the conception stage, he identified the differences of their problem compared with existing studies like car-hailing, i.e., the order dispatching in gig delivery is concurrent, which brings additional challenges. In the system building stage, he designed the time estimation module in the system, and also helped Baoshen Guo (the first author) to decide the appropriate RL model for the system. In the paper writing stage, Yi assisted Baoshen to organize the sections and polish the writing.
"It's an honor that our work has been acknowledged by the real-time system community. Gig delivery and the gig economy are emerging industries and there are many research and practical problems to be studied," shared Ding.