This paper addresses the challenge of estimating local surface differential properties, specifically surface normals and curvatures, from raw 3D point clouds. Traditional methods either rely on fitting pre-defined analytic surfaces risking model bias, or directly regress normals and curvatures overlooking their intrinsic geometric correlation. We propose a learning-based approach that locally fits osculating implicit quadrics to recover both normals and curvatures simultaneously. Drawing on classical differential geometry, we exploit the fact that every point on a C² surface admits an osculating quadric in Monge form that exactly reproduces local differential properties. However, the Monge frame itself depends on the very differential quantities being estimated. To bypass this circularity, we reformulate the Monge-form quadric as an implicit representation in a canonical local frame derived solely from point coordinates, enabling supervised learning without requiring Monge frame alignment. This reformulation allows us to construct a ground-truth dataset of such local-frame quadrics and train a neural network to predict per-point weights and offsets for a robust weighted least squares fitting process. The learned offsets account for the deviations of neighboring points from the idealized osculating surface. We further incorporate stable curvature formulations into the training loss alongside normal supervision to enhance estimation fidelity. Extensive experiments on diverse datasets demonstrate that our method outperforms prior approaches in normal and curvature estimation from raw point clouds.

The orientation of surface normals in 3D point cloud is a fundamental problem in computer vision and graphics. Determining a globally consistent orientation solely from the point cloud is however challenging due to the global scope of the problem and the discrete nature of point cloud, particularly in the presence of noise, outliers, holes, thin structures, and complex topologies. This paper presents an efficient, robust, and global algorithm for generating consistent normal orientation of a dense 3D point cloud. The basic idea is to transform the original binary normal orientation problem to finding a relaxed sign field on a Delaunay graph, which can be achieved by solving a sparse linear system. The Delaunay graph is constructed by triangulating a level set of an implicit function defined from the input point cloud. The shape diameter function is estimated to serve as a prior for determining an appropriate level value such that the level set implicitly defines the inner and outer shells enclosing the input point clouds. As such, our algorithm leverages the strengths of the shape diameter function, Delaunay triangulation, and the least-square techniques, making the underlying processes take both geometry and topology into consideration, and thus provides an efficient and robust solution for handling point clouds with complicated geometry and topology. Extensive experiments on various shapes with noise and outliers confirm the effectiveness and robustness of our algorithm.

This paper introduces ZeroMatch, a novel zero-shot RGB-D point cloud registration framework, aimed at achieving robust 3D matching on unseen data without any task-specific training. Our core idea is to utilize the powerful zero-shot image representation of Stable Diffusion, achieved through extensive pre-training on large-scale data, to enhance point-cloud geometric descriptors for robust matching. Specifically, we combine the handcrafted geometric descriptor FPFH with Stable-Diffusion features to create point descriptors that are both locally and contextually aware, enabling reliable RGB-D registration with zero-shot capability. This approach is based on our observation that Stable-Diffusion features effectively encode discriminative global contextual cues, naturally alleviating the feature ambiguity that FPFH often encounters in scenes with repetitive patterns or low overlap. To further enhance cross-view consistency of Stable-Diffusion features for improved matching, we propose a coupled-image input mode that concatenates the source and target images into a single input, replacing the original single-image mode. This design achieves both inter-image and prompt-to-image consistency attentions, facilitating robust cross-view feature interaction and alignment. Finally, we leverage feature nearest neighbors to construct putative correspondences for hypothesize-and-verify transformation estimation. Extensive experiments on 3DMatch, ScanNet, and ScanLoNet verify the excellent zero-shot matching ability of our method. [Code]

In this paper, we propose a novel 3D registration paradigm, Generative Point Cloud Registration, which bridges advanced 2D generative models with 3D matching tasks to enhance registration performance. Our key idea is to generate cross-view consistent image pairs that are well-aligned with the source and target point clouds, enabling geometric-color feature fusion to facilitate robust matching. To ensure high-quality matching, the generated image pair should feature both 2D-3D geometric consistency and cross-view texture consistency. To achieve this, we introduce Match-ControlNet, a matching-specific, controllable 2D generative model. Specifically, it leverages the depth-conditioned generation capability of ControlNet to produce images that are geometrically aligned with depth maps derived from point clouds, ensuring 2D-3D geometric consistency. Additionally, by incorporating a coupled conditional denoising scheme and coupled prompt guidance, Match-ControlNet further promotes cross-view feature interaction, guiding texture consistency generation. Our generative 3D registration paradigm is general and could be seamlessly integrated into various registration methods to enhance their performance. Extensive experiments on 3DMatch and ScanNet datasets verify the effectiveness of our approach.

Recovering human trajectories from incomplete or missing data is crucial for many mobility-based urban applications, e.g., urban planning, transportation, and location-based services. Existing methods mainly rely on recurrent neural networks or attention mechanisms. Though promising, they encounter limitations in capturing complex spatial-temporal dependencies in low-sampling trajectories. Recently, diffusion models show potential in content generation. However, most of proposed methods are used to generate contents in continuous numerical representations, which cannot be directly adapted to the human location trajectory recovery. In this paper, we introduce a conditional diffusion-based trajectory recovery method, namely, DiffMove. It first transforms locations in trajectories into the embedding space, in which the embedding denoising is performed, and then missing locations are recovered by an embedding decoder. DiffMove not only improves accuracy by introducing high-quality generative methods in the trajectory recovery, but also carefully models the transition, periodicity, and temporal patterns in human mobility. Extensive experiments based on two representative real-world mobility datasets are conducted, and the results show significant improvements (an average of 11% in recall) over the best baselines.

Vehicle trajectory data are essential in smart mobility applications, yet often incomplete, necessitating systematic reconstruction for effective use. Existing methods often overlook traffic rules and vehicle interactions in their reconstruction process, a research gap that becomes critical for fine-grained reconstruction of incomplete and irregular microscopic traffic data. To address this limitation, this paper introduces a novel fine-grained trajectory reconstruction (FTR) framework, particularly for urban signalized intersections, considering both traffic rules and vehicle interactions through a microscopic traffic simulation (MTS) model. This is motivated by challenging missing patterns in real-world data from Alibaba City Brain Lab and limitations in existing reconstruction approaches. To this end, the FTR problem is first formulated as an MTS-based optimization problem. Then, to solve this problem effectively under a limited computing budget, an advanced dynamic data-driven evolutionary optimization technique, D3GA++, is proposed. Through the validation involving two real-world datasets, D3GA++ has demonstrated superior performance under various missing data scenarios consistently surpassing baselines such as brute-force random search and standard evolutionary algorithm in terms of reconstruction accuracy. Our work can have crucial implications for traffic management, urban planning, and autonomous vehicle technology development.

With the increasing popularity of GPS modules, there are various urban applications such as car navigation relying on trajectory data modeling. In this work, we study the problem of modeling on-road trajectories, which is to predict the next road segment given a partial GPS trajectory. Existing methods that model trajectories with Markov chain or recurrent neural network suffer from various issues, including limited capability of sequential modeling, insufficiency of incorporating the road network context, and lack of capturing the underlying semantics of trajectories. In this article, we propose a new trajectory modeling framework called Multi-task Modeling for Trajectories (MMTraj+), which avoids these issues. Specifically, MMTraj+ uses multi-head self-attention networks for sequential modeling, captures the overall road network as the context information for road segment embedding, and performs an auxiliary task of predicting the trajectory destination information (namely the ID and bearing angle) to better guide the main trajectory modeling task (controlled by a carefully designed gating mechanism). In addition, we tailor MMTraj+ for the cases where the destination information is known by dropping its auxiliary task of predicting the trajectory destination information. Extensive experiments conducted on real-world datasets demonstrate the superiority of the proposed method over the baseline methods.

Road extraction is a process of automatically generating road maps mainly from satellite images. Existing models all target to generate roads from the scratch despite that a large quantity of road maps, though incomplete, are publicly available (e.g. those from OpenStreetMap) and can help with road extraction. In this paper, we propose to conduct road extraction based on satellite images and partial road maps, which is new. We then propose a two-branch Partial to Complete Network (P2CNet) for the task, which has two prominent components: Gated Self-Attention Module (GSAM) and Missing Part (MP) loss. GSAM leverages a channel-wise self-attention module and a gate module to capture long-range semantics, filter out useless information, and better fuse the features from two branches. MP loss is derived from the partial road maps, trying to give more attention to the road pixels that do not exist in partial road maps. Extensive experiments are conducted to demonstrate the effectiveness of our model, e.g. P2CNet achieves state-of-the-art performance with theIoUscores of 70.71% and 75.52%, respectively, on the SpaceNet and OSM datasets.

The phenomena of data missing are common in the field of traffic, yet existing solutions for data imputation are not sufficient due to challenges of data sparsity, complex traffic situations and the lack of complete ground truths. In this paper, we propose a novel solution called STCPA for the speed imputation problem. STCPA captures complex traffic correlations among the spatial and temporal dimensions via the attention mechanism, which helps mitigate the data sparsity issue. In addition, STCPA adopts an imputation cycle consistency constraint for providing reliable supervisions on unobserved entries, which improves the training. Furthermore, it incorporates an extra Road-aware Perceptual Loss, which helps encourage to preserve more meaningful semantics for imputation. Extensive experiments are conducted on two real-world datasets, namely, Chengdu and New York, to demonstrate the effectiveness of STCPA, e.g., it outperforms the best baseline by 7.64% and 5.00% on Chengdu and New York datasets, respectively. The code is available at https://github.com/Sam1224/STCPA.

With the increasing popularity of GPS modules, there are various urban applications relying on trajectory data modeling. In this work, we study the problem to model the vehicle trajectories by predicting the next road segment given a partial trajectory. Existing methods that model trajectories with Markov chain or recurrent neural network suffer from issues of modeling, context and semantics. In this paper, we propose a new trajectory modeling framework called Multi-task Modeling for Trajectories (MMTraj), which avoids these issues. Specifically, MMTraj uses multi-head self-attention networks for sequential modeling, captures the overall road network as the context information for road segment embedding, and performs an auxiliary task of predicting the trajectory destination to better guide the main trajectory modeling task (controlled by a carefully designed gating mechanism). Extensive experiments conducted on real-world datasets demonstrate the superiority of the proposed method over the baseline methods.

With the increasing popularity of Digital Twin, there is an opportunity to employ deep learning models in symbiotic simulation system. Symbiotic simulation can replicate multiple what-if simulation instances from its real-time reference simulation (base simulation) for short-term forecasting. Hence, it is a useful tool for just-in-time decision making process. Recent trends on symbiotic simulation studies emphasize on its combination with machine learning. Despite its success and usefulness, very few works focus on application of such a hybrid system in microscopic traffic simulation. Existing application of machine (deep) learning models in microscopic traffic simulation is confined to either predictive analysis or offline simulation-based prescriptive analysis. Thus, there is also lack of work on updating parameters of a deep learning model dynamically for real-time traffic simulation. This is necessary if the learning-based model is to be used as part of the base simulation so that "Just-in-time (JIT)" what-if simulation initialized from the model can make better short-term forecasts. This paper proposes a data-driven modelling and simulation framework to dynamically update parameters of Long Short-term Memory (LSTM) for JIT microscopic traffic simulation. Extensive experiments were carried out to demonstrate its effectiveness in terms of more accurate short-term forecasting than other baseline models.

Vehicular path flow (trajectories) is an important data source for smart mobility, from which many road traffic parameters can be inferred. However, it has been a long-existing challenge that single source of trajectory data is biased in terms of its spatiotemporal coverage. In this paper, we leverage two types of large traffic datasets - point flows and sample trajectories - to generate the full city-scale vehicular paths. Our method consists of a low-granularity data fusion (LGDF) module, which uses point flow data to estimate the sparse paths that pass through some specific links (where sensors are mounted), and a high-granularity model training (HGMT) component, which uses sample trajectory data to pre-train a bi-gram sequence generation model. Afterwards, the results from LGDF and HGMT are combined to produce detailed on-road spatiotemporal paths. In this way, the data safety of single trajectory is protected while the full-scale city traffic can be reproduced for transportation analytics. The proposed method is verified via real-data case studies. As a result, starting from August 2019, this method has been implemented in Alibaba's city brain project and successively deployed in many cities in China for the purpose of traffic analysis and optimization.

The response time of emergency vehicles (EV) is critical to saving lives. In this paper, we address the two challenges to the reduction of EV response time: 1) path-searching for a reliable estimated time of arrival (ETA) in ever-changing traffic conditions and 2) elastic signal preemption to reduce the negative impact on the whole traffic flow introduced by prioritizing the EVs. While traditional path-searching methods aim to minimize only the mean value of ETA, our approach minimizes a combination of its mean and variance, as we found that the variance of speed has a significant impact on path searching. The simulation shows that our approach generates paths with reliable ETA, i.e., lower variance with the same level of accuracy. Furthermore, we formulate the elastic signal preemption (ESP) as a set of quadratic programming (QP) problems to find non-intrusive signal schedules to the fast track EVs through junctions without stopping. When the ESP module was applied in the real world, results showed about 30% reduction of response time with little impact on the overall traffic conditions.

Traffic signal is the key method for city traffic control. Existing signal control systems use the loop detector data as the main input which is nearsighted in terms of both space and time. Lacking of effective data collection methods has hindered the development of more sophisticated models. It is therefore very hard to develop an optimization model considering all signals in a region or even a city. In this paper, we will introduce our method for large scale traffic signal optimization, which is the major module of Alibaba's city brain solution. By integrating multiple data sources to sense the whole city's traffic conditions, a layered model is developed based on the divide-and-conquer paradigm to gradually apply different types of data-driven optimization algorithms. It is a paradigm shift to use big data to improve a traditionally closed signal control system, and its effectiveness has been proven in a field test in Shanghai city.

Public bus services are often planned in the context of urban planning. For a city with efficient and extensive network of public transportation system like Singapore, enhancing the existing coverage of bus service to meet the dynamic mobility needs of the population requires data mining approach. Specifically, frequent taxi rides between two locations at a period of time may suggest possible poor coverage of public transport service, if not lacking of the public transport service. In this paper, we describe a proof of concept effort to discover this weakness and its improvement in public transportation system via mining of taxi ride dataset. We cluster taxi rides dataset to determine some popular taxi rides in Singapore. From the clustered taxi rides, we filter and select only the clusters whose commuting via existing public transport are tortuous if not unreachable door-to-door. Based on the discovered travel pattern, we propose new bus routes that serve the passengers of these clusters. We formulate the bus planning problem as an optimization of directed cycle graph, and present it's preliminary solution and results. We showcase our idea in the case of Singapore.