DRL4AOI: A DRL Framework for Semantic-aware AOI Segmentation in Location-Based Services

TL;DR

DRL4AOI reframes AOI segmentation in Location-Based Services as a Markov Decision Process and employs deep reinforcement learning to optimize service-oriented rewards. The TrajRL4AOI specialization for logistics uses trajectory modularity and road-network alignment as dual objectives via a Double-DQN, achieving superior partition quality on both synthetic and real-world data. A public codebase and an AOI segmentation system support practical deployment and evaluation. Overall, the framework provides a flexible, scalable approach to integrate geo- and service-related information for semantic-aware AOI generation in LBS.

Abstract

In Location-Based Services (LBS), such as food delivery, a fundamental task is segmenting Areas of Interest (AOIs), aiming at partitioning the urban geographical spaces into non-overlapping regions. Traditional AOI segmentation algorithms primarily rely on road networks to partition urban areas. While promising in modeling the geo-semantics, road network-based models overlooked the service-semantic goals (e.g., workload equality) in LBS service. In this paper, we point out that the AOI segmentation problem can be naturally formulated as a Markov Decision Process (MDP), which gradually chooses a nearby AOI for each grid in the current AOI's border. Based on the MDP, we present the first attempt to generalize Deep Reinforcement Learning (DRL) for AOI segmentation, leading to a novel DRL-based framework called DRL4AOI. The DRL4AOI framework introduces different service-semantic goals in a flexible way by treating them as rewards that guide the AOI generation. To evaluate the effectiveness of DRL4AOI, we develop and release an AOI segmentation system. We also present a representative implementation of DRL4AOI - TrajRL4AOI - for AOI segmentation in the logistics service. It introduces a Double Deep Q-learning Network (DDQN) to gradually optimize the AOI generation for two specific semantic goals: i) trajectory modularity, i.e., maximize tightness of the trajectory connections within an AOI and the sparsity of connections between AOIs, ii) matchness with the road network, i.e., maximizing the matchness between AOIs and the road network. Quantitative and qualitative experiments conducted on synthetic and real-world data demonstrate the effectiveness and superiority of our method. The code and system is publicly available at https://github.com/Kogler7/AoiOpt.

Figures (12)

• Figure 1: Illustration of Different Methods.
• Figure 2: Illustration of Action. It is defined as merging the target grid to its up/down/left/right AOI or staying in its current AOI.
• Figure 3: DRL4AOI Architecture, which mainly includes three steps: 1) Data preprocessing, which rasterizes the map and prepares service-related input. 2) RL-based AOI segmentation, which gradually adjusts the grids in AOI's border by the agent under the guidance of the rewards designed by service-semantic goals. 3) Post-processing, which is designed to make further refinement to the segmentation results.
• Figure 4: Motivation of introducing trajectory. Two communities (i.e., Com1 and Com2) are surrounded by the same road network, separated by a fence. It is unreasonable to divide them into the same AOI according to the road network. By leveraging a vast amount of courier trajectory data, we can better distinguish between Com1 and Com2, resulting in more accurate and suitable AOI segmentations.
• Figure 5: AOI initialization based on the road network. It contains two steps: 1) obtain the road network data and export it as an image. 2) perform binary thresholding, connected components, and contour extraction and expansion.