A Spatial-Temporal Attention Model for Human Trajectory Prediction

Human trajectory prediction is essential and promising in many related applications. This is challenging due to the uncertainty of human behaviors, which can be influenced not only by himself, but also by the surrounding environment. Recent works based on long-short term memory(LSTM) models have brought tremendous improvements on the task of trajectory prediction. However, most of them focus on the spatial influence of humans but ignore the temporal influence. In this paper, we propose a novel spatial-temporal attention(ST-Attention) model,which studies spatial and temporal affinities jointly. Specifically,we introduce an attention mechanism to extract temporal affinity,learning the importance for historical trajectory information at different time instants. To explore spatial affinity, a deep neural network is employed to measure different importance of the neighbors. Experimental results show that our method achieves competitive performance compared with state-of-the-art methods on publicly available datasets.

Frequent Trajectory Patterns Mining for Intelligent Visual Surveillance System

A frequent trajectory patterns mining algorithm is proposed to learn the object activities and classify the trajectories in intelligent visual surveillance system.The distribution patterns of the trajectories were generated by an Apriori based frequent patterns mining algorithm and the trajectories were classified by the frequent trajectory patterns generated.In addition,a fuzzy c-means(FCM)based learning algorithm and a mean shift based clustering procedure were used to construct the representation of trajectories.The algorithm can be further used to describe activities and identify anomalies.The experiments on two real scenes show that the algorithm is effective.

An Algorithm for Mining Gradual Moving Object Clusters Pattern From Trajectory Streams

The discovery of gradual moving object clusters pattern from trajectory streams allows characterizing movement behavior in real time environment,which leverages new applications and services.Since the trajectory streams is rapidly evolving,continuously created and cannot be stored indefinitely in memory,the existing approaches designed on static trajectory datasets are not suitable for discovering gradual moving object clusters pattern from trajectory streams.This paper proposes a novel algorithm of gradual moving object clusters pattern discovery from trajectory streams using sliding window models.By processing the trajectory data in current window,the mining algorithm can capture the trend and evolution of moving object clusters pattern.Firstly,the density peaks clustering algorithm is exploited to identify clusters of different snapshots.The stable relationship between relatively few moving objects is used to improve the clustering efficiency.Then,by intersecting clusters from different snapshots,the gradual moving object clusters pattern is updated.The relationship of clusters between adjacent snapshots and the gradual property are utilized to accelerate updating process.Finally,experiment results on two real datasets demonstrate that our algorithm is effective and efficient.

Characteristics of evolution of mining-induced stress field in the longwall panel:insights from physical modeling

The evolution of mining-induced stress field in longwall panel is closely related to the fracture field and the breaking characteristics of strata.Few laboratory experiments have been conducted to investigate the stress field.This study investigated its evolution by constructing a large-scale physical model according to the in situ conditions of the longwall panel.Theoretical analysis was used to reveal the mechanism of stress distribution in the overburden.The modelling results showed that:(1)The major principal stress field is arch-shaped,and the strata overlying both the solid zones and gob constitute a series of coordinated load-bearing structures.The stress increasing zone is like a macro stress arch.High stress is especially concentrated on both shoulders of the arch-shaped structure.The stress concentration of the solid zone in front of the gob is higher than the rear solid zone.(2)The characteristics of the vertical stress field in different regions are significantly different.Stress decreases in the zone above the gob and increases in solid zones on both sides of it.The mechanical analysis show that for a given stratum,the trajectories of principal stress are arch-shaped or inverselyarched,referred to as the‘‘principal stress arch’’,irrespective of its initial breaking or periodic breaking,and determines the fracture morphology.That is,the trajectories of tensile principal stress are inversely arched before the first breaking of the strata,and cause the breaking lines to resemble an inverted funnel.In case of periodic breaking,the breaking line forms an obtuse angle with the advancing direction of the panel.Good agreement was obtained between the results of physical modeling and the theoretical analysis.

CLEAN:Frequent Pattern-Based Trajectory Compression and Computation on Road Networks

The volume of trajectory data has become tremendously huge in recent years. How to effectively and efficiently maintain and compute such trajectory data has become a challenging task. In this paper, we propose a trajectory spatial and temporal compression framework, namely CLEAN. The key of spatial compression is to mine meaningful trajectory frequent patterns on road network. By treating the mined patterns as dictionary items, the long trajectories have the chance to be encoded by shorter paths, thus leading to smaller space cost. And an error-bounded temporal compression is carefully designed on top of the identified spatial patterns for much low space cost. Meanwhile, the patterns are also utilized to improve the performance of two trajectory applications, range query and clustering, without decompression overhead. Extensive experiments on real trajectory datasets validate that CLEAN significantly outperforms existing state-of-art approaches in terms of spatial-temporal compression and trajectory applications.

Investigating spatial and temporal variations of soil moisture content in an arid mining area using an improved thermal inertia model

Mining operations can usually lead to environmental deteriorations. Underground mining activities could cause an extensive decrease in groundwater level and thus a dramatic variation in soil moisture content（SMC）. In this study, the spatial and temporal variations of SMC from 2001 to 2015 at two spatial scales（i.e., the Shendong coal mining area and the Daliuta Coal Mine） were analyzed using an improved thermal inertia model with a long-term series of Landsat TM/OLI（TM=Thematic Mapper and OLI=Operational Land Imager） data. Our results show that at large spatial scale（the Shendong coal mining area）, underground mining activities had insignificant negative impacts on SMC and that at small spatial scale（the Daliuta Coal Mine）, underground mining activities had significant negative impacts on SMC. Trend analysis of SMC demonstrated that areas with decreasing trend of SMC were mainly distributed in the mined area, indicating that underground mining is a primary cause for the drying trend in the mining region in this arid environment.

Blockage of the Deep-Sea Mining Pump Transporting Large Particles with Different Sphericity

The present study aims to plumb blockage of the deep-sea mining pump transporting large particles with different shapes. A numerical work was performed through combining the computational fluid dynamics(CFD) technique and the discrete element method(DEM). Six particle shapes with sphericity ranging from 0.67 to 1.0 were selected. A velocity triangle is built with the absolute, relative, and circumferential velocities of particles. Velocity triangles with absolute velocity angles ranging from 90° to 180° prevail in the first-stage impeller. With declining sphericity, more particles follow the velocity triangle with absolute velocity angles ranging from 0° to 90°, which weakens the ability of particles to pass through the flow passage. Furthermore, the forces acting on the particles traveling in the impeller passage are analyzed. Large particles, especially non-spherical ones, suffer from high centrifugal force and therefore move along the suction surface of the impeller blades. Non-spherical particles undergo great drag force as a result of large surface area. The distribution of drag force angles is featured by two peaks, and one vanishes due to blockage.As particle sphericity declines, both magnitude and angle of the pressure gradient force decrease. Variation of the drag force and the pressure gradient force causes clockwise deflection of the centripetal force, resulting in deflection and elongation of particle trajectory, which increases the possibility of blockage.

A System for Detecting Refueling Behavior along Freight Trajectories and Recommending Refueling Alternatives

Smart refueling can reduce costs and lower the possibility of an emergency. Refueling intelligence can only be obtained by mining historical refueling behaviors from big data, however, without devices, such as fuel tank cursors, and cooperation from drivers, these behaviors are hard to detect. Thus, detecting refueling behaviors from big dala derived from easy-to-approach trajectories is one of/he most efficient retrieve evidences for research of refueling behaviors. In this paper, we describe a complete procecdure for detecting refoeling behavior in big data derived from freight trajectories. This procedure involves the inte- gration of spatial data mining and machine-learning techniques. The key pall of the methodology is a pattern detector that extends the naive Bayes classifier. By draw＇ing on the spatial and temporal characteristics of freight trajectories, refileling behaviors can be identified with high accuracy. Fu,lher, we present a refueling prediction and recommendation system to show how our refueling detector can be used practically in big data. Our experimetlts on real trajeclories show that our refueling detector is accurate, and the system performs well.

智能采煤机器人关键技术

采煤机是综采工作面的核心装备,研发智能采煤机器人是实现综采工作面智能化的关键。综合分析当前采煤机机器人化研究进程中的传感检测、位姿控制、速度控制、截割轨迹规划与跟踪控制等技术的研究现状,提出研发智能采煤机器人必须破解的“智能感知、位姿控制、速度控制、截割轨迹规划与跟踪控制、位-姿-速协同控制”五大关键技术,并给出解决方案。针对智能感知问题,提出了构建智能感知系统思路,给出了智能采煤机器人智能感知系统的架构,实现对运行状态、位姿、环境等全面感知,为智能采煤机器人安全、可靠运行提供保障;针对位姿控制问题,提出了智能PID位姿控制思路,给出了改进遗传算法的PID位姿控制方法,实现了智能采煤机器人位姿精准控制;针对速度控制问题,提出了融合“力-电”异构数据的截割载荷测量思路,给出了基于神经网络算法的截割载荷测量方法,实现了截割载荷的精准测量;提出牵引与截割速度自适应控制思路,给出了人工智能算法牵引与截割速度决策方法和滑模自抗扰控制的牵引与截割速度控制方法,实现了智能采煤机器人速度精准自适应控制;针对截割轨迹规划与跟踪控制问题,提出了截割轨迹精准规划思路,给出了融合地质数据和历史截割数据的截割轨迹规划模型,实现了截割轨迹的精准规划;提出了截割轨迹精准跟踪控制思路,给出了智能插补算法的截割轨迹跟踪控制方法,实现了智能采煤机器人截割轨迹高精度规划与精准跟踪控制;针对“位-姿-速”协同控制问题,提出了“位-姿-速”协同控制参数智能优化思路,给出了基于多系统互约束的改进粒子群“位-姿-速”协同控制参数优化方法,实现了智能采煤机器人智能高效作业。深入研究五大关键技术破解思路,有利于加快推动研发高性能、高效率、高可靠的智能采煤机器人。

Extracting Campus’Road Network from Walking GPS Trajectories

Road network extraction is vital to both vehicle navigation and road planning.Existing approaches focus on mining urban trunk roads from GPS trajectories of floating cars.However,path extraction,which plays an important role in earthquake relief and village tour,is always ignored.Addressing this issue,we propose a novel approach of extracting campus’road network from walking GPS trajectories.It consists of data preprocessing and road centerline generation.The patrolling GPS trajectories,collected at Hunan University of Science and Technology,were used as the experimental data.The experimental evaluation results show that our approach is able to effectively and accurately extract both campus’trunk roads and paths.The coverage rate is 96.21%while the error rate is 3.26%.

基于知识的目标关系分析挖掘技术

在战场目标价值分析和打击目标排序分析过程中,为了构建敌方作战目标体系,需要分析战场目标间的关联关系。轨迹和部署数据中隐藏大量信息,提出了一种从轨迹和部署数据中挖掘出感兴趣的目标关系类型信息的方法,所提方法对轨迹部署数据进行时空聚类,从聚类结果提取目标;对聚类目标使用频繁项挖掘算法分析挖掘满足一定支持度的有关联关系的目标,再根据构建的关系类型知识库或关系规则,分析目标间的具体关系类型。所提方法能对积累的目标历史轨迹部署数据分析挖掘出目标间的关联关系,挖掘出目标潜在的关系类型可为后续构建目标体系提供关系数据。

变结构井下救援机器人动力学建模与轨迹跟踪控制

针对钻孔救援需求,设计了一种变结构井下救援机器人,该机器人可改变自身构型,通过钻孔进入井下环境。根据机器人结构特点,分析了机器人的转向机制,建立了考虑机器人前轮/履带-土壤间力学作用、描述机器人转向特性的动力学模型;为提高机器人的转向性能,分析了系统的瞬态及稳态响应以及不同结构参数对转向性能的影响,并对机器人结构参数进行了改进;基于机器人动力学模型,设计模型预测控制(Model Predictive Control,MPC)控制器,实现了直线轨迹和圆形轨迹的轨迹跟踪仿真。结果表明,所建动力学模型是有效的,MPC轨迹跟踪控制器能够稳定地跟踪期望轨迹,为机器人的控制奠定了基础。

基于频繁序列挖掘的出租车轨迹特性分析

为进一步厘清不同出租车路径选择行为的差异性,采用频繁序列挖掘方法提取了同一个OD对间的频繁路径,构建路径选择集,分别从静态和动态两个角度分析路径集的相似特性。以西安市出租车的轨迹数据为研究对象,通过栅格划分与路网匹配,获得了不同OD对之间的路径集合。重新定义了频繁路径,采用PrefixSpan演变算法,在得到频繁子序列的基础上引入动态阈值和频繁度指标挖掘频繁路径,提取了最短路径和其他路径,完成了3类有效路径集的构建,并分析了路径集的一般属性。其后,将路径上二维时间序列(轨迹)间的相似度表示为动态相似度,将一维有向序列(路段)间的相似度表示为静态相似度,基于改进的最长公共子序列和动态时间规整算法对3类路径进行了相似性分析。结果表明:频繁路径与最短路径的相似度较高,意味着大多数出租车仍然选择具有最低出行时间的路段,但不一定会选择最短路径;时间和距离仍是出行者选择路径时主要考虑的因素,但出行者并不完全追求时间最短或距离最短;试验得到的动态相似度计算结果显著高于静态相似度计算结果,说明路径上的二维时序相似度高于一维形状相似度;两种方法下频繁路径和最短路径的相似度均最高,最短路径和其他路径的相似度均最低,比较结果的一致性说明可以用动态轨迹的相似度来大致度量静态路径的相似度。文中的频繁路径挖掘算法具有一定的可靠性,可为城市交通管理者进行路径推荐、道路规划等提供支持。

基于手势识别的矿车智能控制技术研究

针对露天矿区用于物料运输的矿车控制需求,设计一种基于手势识别的矿车智能控制系统。首先,通过配有加速度传感器的数据手套采集手势数据,经蓝牙模块实现信息的无线传输,再利用微控制器ATmega328P实现控制信息的分析和处理,采用PID控制算法得出矿车的电机转动参数,从而实现对矿车车轮的控制。同时,设计以树莓派4B为主控,搭载有LCD显示器、扬声器、麦克风和摄像头等模块的系统,用以实现音视频的实时互传等功能。实验结果表明,该系统能根据手势的变化对矿车的行驶进行准确地控制,具有十分可观的应用前景。

轨迹异常检测研究综述

传感器技术的飞速发展催生大量交通轨迹数据,轨迹异常检测在智慧交通、自动驾驶、视频监控等领域具有重要的应用价值.不同于分类、聚类和预测等轨迹挖掘任务,轨迹异常检测旨在发现小概率、不确定和罕见的轨迹行为.轨迹异常检测中一些常见的挑战与异常值类型、轨迹数据标签、检测准确率以及计算复杂度有关.针对上述问题,全面综述近20年来轨迹异常检测技术的研究现状和最新进展.首先,对轨迹异常检测问题的特点与目前存在的研究挑战进行剖析.然后,基于轨迹标签的可用性、异常检测算法原理、离线或在线算法工作方式等分类标准,对现有轨迹异常检测算法进行对比分析.对于每一类异常检测技术,从算法原理、代表性方法、复杂度分析以及算法优缺点等方面进行详细总结与剖析.接着,讨论开源的轨迹数据集、常用的异常检测评估方法以及异常检测工具.在此基础上,给出轨迹异常检测系统架构,形成从轨迹数据采集到异常检测应用等一系列相对完备的轨迹挖掘流程.最后,总结轨迹异常检测领域关键的开放性问题,并展望未来的研究趋势和解决思路.

矿用无人运输车辆轨迹跟踪控制算法研究

矿用无人运输车辆作业环境恶劣,存在大曲率弯道、坡道等非结构化道路明显特征,对无人化运输控制要求高。为改善PID等传统控制算法适应性问题,提高无人驾驶轨迹跟踪的车辆横纵向控制精度,提出一种纯跟踪与PID结合的多点预瞄横向控制、考虑模糊控制表参数拟合的纵向控制方法,减少控制参数的同时提高算法效果。根据传统控制算法设计基础控制器,结合基础算法优势进行横向与纵向控制算法设计,通过硬件在环仿真和实车测试验证算法的性能。试验结果表明,横向控制算法与斯坦利算法相比,车辆路径跟踪精度有明显改善,纵向控制方面,速度跟随误差<1 km/h,保证了车辆驾驶时的平稳性与舒适性。

基于轨迹数据的大规模路网交通拥挤时空关联规则挖掘

提出了K近邻RElim(K neighbor-RElim,KNR)算法和时序K近邻RElim(sequential KNbrRElim,SKNR)算法,利用大规模路网的车辆轨迹数据来挖掘路段拥挤关联规则和拥挤传播时空关联规则。其中KNR算法在RElim算法基础上拓展了空间拓扑约束,可高效从大规模车辆轨迹数据集中挖掘路网中关联性拥挤易发路段,并量化这些路段间拥挤的关联性强度。而SKNR算法进一步以滑动窗口的形式拓展时间维度,可以挖掘出大规模路网中难以直接观测的拥挤传播现象,并追溯拥挤传播路径。以成都路网和车辆轨迹数据的挖掘结果对所提出的算法进行了说明和验证,结果表明了算法的有效性和鲁棒性。

金属矿山井下采场六足机器人运动分析及步态规划

六足机器人因其结构特殊带来的良好越障能力成为仿生机器人研究的热点,然而,金属矿山井下采场矿石堆积、崎岖不平的特点,给这类机器人的行走稳定性、越障性带来更多的挑战。因此,为使六足机器人在井下具有更好的通过性,对其运动能力和步态规划进行了相关研究。首先仿照自然界六足生物,设计六足机器人结构,对其腿部进行运动学分析;然后规划了用于采场的直行步态,结合采场路面环境设计了一种直线—摆线复合轨迹提升越障性,同时分析了机器人爬坡稳定性,对爬坡步态进行了优选;最后对规划步态进行仿真和现场模拟试验。仿真和试验结果表明,所规划的足端轨迹能跨越抬腿高度85%的障碍物,并且对矿石堆积形成的采场路面有更好的避障能力;三角步态爬坡时在坡底和坡顶过渡阶段容易打滑,而横向步态可以实现平滑的过渡,爬坡性能更佳。