Online food delivery (OFD) services, known for offering varied meals at home, have gained global popularity. Meituan has recently ventured into the affordable market segment with its "Pinhaofan'' service, highlighting the imperative to delivery efficiency. To achieve this, delivery scope is regarded as one of the most effective operational tools. The delivery scope of a merchant refers to the geo-graphical area where they can serve customers. Current methods for generating delivery scopes primarily focus on optimizing a single merchant's efficiency or rely on manual delineated from the merchant's perspective, neglecting the merchant substitution effect and potentially resulting in order loss. In this paper, we propose a novel method, named Collaborative Scope, which views the delivery scope as an assortment optimization problem, considering the substitution effect between merchants from the user's perspective. We introduce the discrete choice model of econometrics and use the Enhanced Multinomial Logit Model to predict user conversion rates in the merchant list. Next, we formulate the delivery scope optimization problem of multiple merchants as a mixed integer programming problem. The city-wide solution of this problem, owing to the large-scale combinatorial optimization triggered by high-dimensional decision variables, incurs high computational complexity. To address this, we propose an approximate solution to the original problem through a first-order Taylor series approximation, which significantly reduces the computation complexity at the expense of a slight decrease in solution accuracy. Offline and online A/B test results indicate that, compared to existing methods, Collaborative Scope significantly improves delivery efficiency by reducing delivery difficulty without hurt of order volume. Notably, Collaborative Scope is currently deployed on "Pinhaofan'', serving tens of millions of online users.

With the proliferation of intelligent sensors integrated into mobile devices, fine-grained human activity recognition (HAR) based on lightweight sensors has emerged as a useful tool for personalized applications. Although shallow and deep learning algorithms have been proposed for HAR problems in the past decades, these methods have limited capability to exploit semantic features from multiple sensor types. To address this limitation, we propose a novel HAR framework, DiamondNet, which can create heterogeneous multisensor modalities, denoise, extract, and fuse features from a fresh perspective. In DiamondNet, we leverage multiple 1-D convolutional denoising autoencoders (1-D-CDAEs) to extract robust encoder features. We further introduce an attention-based graph convolutional network to construct new heterogeneous multisensor modalities, which adaptively exploit the potential relationship between different sensors. Moreover, the proposed attentive fusion subnet, which jointly employs a global-attention mechanism and shallow features, effectively calibrates different-level features of multiple sensor modalities. This approach amplifies informative features and provides a comprehensive and robust perception for HAR. The efficacy of the DiamondNet framework is validated on three public datasets. The experimental results demonstrate that our proposed DiamondNet outperforms other state-of-the-art baselines, achieving remarkable and consistent accuracy improvements. Overall, our work introduces a new perspective on HAR, leveraging the power of multiple sensor modalities and attention mechanisms to significantly improve the performance.

Online food delivery (OFD) services have become popular globally, serving people's daily needs. Precise area-of-interest (AOI) boundaries help OFD platforms determine customers' exact locations, which is crucial for maintaining consistency in delivery difficulty and providing a uniform customer experience within an AOI. Existing AOI generation methods primarily rely on predefined shapes or density-based clustering, which limits the quality of the contours. Recently, Meituan has treated the AOI contours as a binary semantic segmentation problem. Their approach involves a multi-step post-process to address the issues with boundary breaks caused by semantic segmentation models, leading to decreased quality and inefficiency in the learning process. In this paper, we propose a novel method for AOI contour generation called C-AOI (Contour-based Area-of-Interest). C-AOI is an instance segmentation model that focuses on generating high-quality AOI contours. Unlike the former method, which relies on pixel-by-pixel classification, C-AOI starts from the center point of the AOI and regresses the boundary. This approach results in a higher-quality boundary and is less computationally intensive. C-AOI first corrects errors on the contour using a local aggregation mechanism. Then, we propose a novel deforming module called the contour transformer, which captures the global geometry of the object. To enhance the positional relationship among vertices, we introduce a learnable cyclic positional encoding applied to the contour transformer. Finally, to improve the boundary details, we propose the Adaptive Matching Loss (AML) that eliminates over-smoothed boundaries and promotes optimized convergence pathways. Experimental results on real-world datasets collected from Meituan have demonstrated that C-AOI significantly improves the mask and boundary quality compared to Meituan's previous work. Moreover, Its inference speed is comparable to that of E2EC, a state-of-the-art real-time contour-based method. It is noteworthy that C-AOI has been deployed in the Meituan platform for producing AOIs.

Human activity recognition (HAR) based on wearable sensors has been a prosperous research topic in recent years. Considering that the sensor may be worn at diverse body positions, obtaining enough labeled human activity data for each body-worn position is usually expensive and labor-intensive. Furthermore, the variability and diversity of data distribution induced by different body-worn positions make the HAR model trained on data collected from one body position perform poorly for other body-worn positions. To achieve accurate HAR with low labeling cost, which we call unsupervised cross-position HAR, in this article, we propose a deep multiscale transfer learning (DMSTL) model. In the model, we first introduce an unsupervised source selection method to select the most similar source domain for transferring domain knowledge. Then, we develop a multiscale spatial–temporal Net (MSSTNet) to learn comprehensive multimodal representations from multiple feature subspaces. Finally, we design the category-level adaptation module and the domain-level adversarial module for learning domain-invariant features. We conduct extensive experiments on three public HAR data sets and demonstrate the reasonable generalization performance of the DMSTL, which remarkably outperforms other state-of-the-art baselines.

Accurate location data of ground vehicles is very important for various intelligent transportation applications. As one of the most commonly-used navigation solutions at present, the GNSS/INS integrated navigation still cannot meet the accuracy and stability command of current applications. This paper presents a National Marine Electronics Association (NMEA) protocol data-assisted high-accuracy navigation algorithm based on GNSS position error estimation using multi-task learning (MT-e&R), which can accurately estimate the GNSS position error and GNSS measurement noise covariance matrix with the assistance of protocol data and a multi-task learning model. Extensive experimental results on practical navigation data collected in various urban environments of Beijing demonstrate that our proposed approach can improve the performance of integrated navigation system. The positioning errors of integrated navigation equipped with single-frequency receiver are reduced by 36.17% and 39.58% for double-frequency receiver, which confirms the reasonable environmental adaptability of our proposed MT-e&R algorithm.

With the ever-increasing intelligent perception capability of mobile terminals, the demand for fine-grained human activity recognition has become urgent. Transportation mode recognition, as a special branch of human activity recognition, plays a vital role in various mobile smart services. Some studies have been conducted on attitude-independence transportation mode recognition from various perspectives. However, it is challenging to recognize cross-place transportation mode due to the diversity of human activity and carrier deployment place. Towards this end, we propose a robust cross-place transportation mode recognition algorithm, which consists of three parts: Multi-Sensor Neural Network model, a variant bootstrap (ensemble learning) method, and data augmentation. The Multi-Sensor Neural Network model leverages multiple SF-SEDNets to extract spatial-temporal and spectral fusion features from different sensor combinations. The proposed data augmentation method addresses the unbalance data problem without introducing additional data collection costs, and the variant bootstrap method improves the robustness of our proposed algorithm. We evaluate our proposed algorithm on the Sussex-Huawei Locomotion-Transportation recognition challenge 2019 dataset. Extensive experimental results indicate that our algorithm achieves 81.45% macro-F1 score on the test dataset, which is 3.03% higher than that provided by the winning entry of this competition and is 14.85% higher than that of the official baseline.

In recent years, with the widespread popularity of smart mobile devices, seamless location-based service (LBS) between indoor and outdoor (IO) space has become an emerging requirement. As the direct indicator of location and contextual status, accurate indoor/outdoor status guarantees stability and continuity for human localization and activity recognition. Rich-sensor smartphones have made possible various context sensing. However, it is challenging to realize fast and accurate IO switching detection in dynamic and complex environments with variable sensory signals. Toward this end, we propose a multisensor deep learning model to predict the IO state, which consists of four parts: 1) sensor vectors; 2) 1-D sensor DenseNet; 3) LSTM; and 4) MLP. Unlike the traditional feature engineering and fixed threshold strategy, we utilize 1-D sensor DenseNet to extract higher level features for different sensors, and then utilize LSTM to capture signal change pattern in time series. Finally, we leverage the MLP to classify indoor/outdoor scenes. We evaluated our proposed architecture on two data sets, one for only differentiating IO scenes and one for detecting the indoor/outdoor switching scenes. Extensive experimental results show that our proposed method outperforms the fixed threshold method, traditional machine learning-based methods, and other existing neural network-based methods in terms of switching delay and recognition accuracy.