Papers

Papers Fulfillment & Last-Mile

How Meituan Uses Gaussian Mixture Models to Optimize Food Delivery: New Research from Tsinghua University

Tsinghua University and Meituan collaborate on research using Gaussian Mixture Models to model stochastic service times in food delivery, achieving efficient solutions through a hybrid estimation of distribution algorithm and validating practical value through real-world A/B testing.

2026-02-27

Papers Fulfillment & Last-Mile

Drone-Rider Collaborative Delivery: HKUST Research Reveals Infrastructure Planning for Urban Low-Altitude Economy

HKUST research proposes drone-rider collaborative delivery model, solving MINLP through neural network-assisted optimization, achieving 23.5% cost reduction in Shenzhen case, providing scientific basis for urban low-altitude economy infrastructure planning.

2026-02-27

Papers Fulfillment & Last-Mile

Georgia Tech × Grubhub’s Landmark 70-Page Paper: Defining the Meal Delivery Routing Problem (MDRP)

Georgia Tech and Grubhub's foundational paper formally defines MDRP, proposes a two-stage commitment framework and priority scheduling, establishing the research foundation for the entire meal delivery optimization field.

2026-02-19

Papers Fulfillment & Last-Mile

KDD Paper in Production: Tsinghua × Meituan Learn Order Pooling from Skilled Couriers, 55% Peak Efficiency Gain

Meituan × Tsinghua KDD paper proposes SCDN framework to mine order pooling knowledge from skilled courier trajectories. Deployed in production, achieving 45-55% courier efficiency gain during lunch peak while maintaining on-time delivery.

2026-02-18

Papers Fulfillment & Last-Mile

Transportation Science Paper: RL + Hyper-Heuristics Cut Meituan Meal Delivery Costs by 12%

A frontier paper submitted to Transportation Science proposes an n-step SARSA and MAB hyper-heuristic hybrid framework, validated on Meituan's real data. Key finding: strategic order postponement cuts costs 12%; 10% more couriers beats algorithm improvements.

2026-02-18

Papers Fulfillment & Last-Mile

Tsinghua’s Landmark Survey: How Deep Reinforcement Learning Is Reshaping Intelligent Dispatch in Logistics and Transportation

A 41-page ACM Computing Surveys paper from Tsinghua, UIUC, and Cambridge proposes the DDS Service Loop framework and comprehensively reviews DRL applications in food delivery, ride-hailing, express shipping, and warehousing.

2026-02-18