It is vital that autonomous vehicles acquire accurate and real-time information about objects in their vicinity, which fully guarantees the safety of passengers vehicle various environments. Three-dimensional light detection ranging (3D LIDAR) sensors can directly obtain position geometric structure an object within its range, whereas use vision cameras most suitable for recognition. Accordingly, this paper, we present a novel identification method fuses complementary obtained by two types...

Effectively detecting road boundaries in real time is critical to the applications of autonomous vehicles, such as vehicle localization, path planning, and environmental understanding. To precisely extract irregular or those blocked by obstructions on from 3D LiDAR data, a dedicated algorithm consisting four steps proposed this paper. The are follows. First, data pre-processed, employing position attitude information, many noise points deleted. Second, ground quickly separated pre-processed...

Graph Convolutional Networks (GCNs) are widely used for skeleton-based action recognition and achieved remarkable performance. Due to the locality of graph convolution, GCNs can only utilize short-range node dependencies but fail model long-range relationships. In addition, existing convolution based methods normally use a uniform skeleton topology all frames, which limits ability feature learning. To address these issues, we present Convolution Network with Self-Attention (SelfGCN),...

Commercial jerky counterfeiting is widespread in the market. This study combined visible-near-infrared and short-wave-near-infrared hyperspectral imaging along with multiple machine learning algorithms for non-destructive identification of five types commercial products, explored impact different spectral bands, algorithm selection, optimization methods on performance. After data preprocessing, all models' accuracies stability improved. Specifically, logistic regression model was best...

This paper formulates the active suspension control problem as disturbance attenuation with output and constraints. The H∞ performance is used to measure ride comfort such that more general road disturbances can be considered, while time-domain hard constraints are captured using concept of reachable sets state-space ellipsoids. Hence, conflicting requirements specified separately handled in a nature way. In framework Linear Matrix Inequality (LMI) optimization, constrained suspensions...

Vehicle platooning has been a major research topic in recent years because of its ability to reduce fuel consumption, enhance road traffic safety and utilize the more efficiently. A practical applicable platoon merging maneuver is key forming new platoons while ensuring economy. This study proposes strategies that consider both safe space acceleration limitations for two adjacent comprising connected autonomous vehicles (CAVs). The distributed model predictive control (DMPC) algorithm...

The generalized /spl Hscr//sub 2/ norm is adapted to capture requirements of satisfying time-domain hard constraints such as suspension stroke limitation, road holding constraint and actuator saturation. peak amplitudes can then be bounded directly. used a measure for ride comfort. This way, conflicting control objectives are specified independently in the framework multi-objective solved using LMI optimization. Simulation results 2 DOF quarter-car model confirm significant improvement on...

This paper exploits the multiobjective control technique to design an output-feedback active suspension. The ride performance is characterized by /spl Hscr/;/sub 2/ norm, while requirements for having good handling and respecting structural constraints actuator saturation are specified generalized norm. suspension problem then formulated in a mixed 2//generalized problem, solution derived using LMI optimization. Simulation results half-car model discussed.

Rain and snowfall will increase noise, change the resolution of objects in point cloud present great challenges to accurate recognition traffic objects. Accordingly, this article proposes an efficient real-time method for roadside 3-D light detection ranging (LIDAR) background extraction object segmentation under snowy weather. We first use a historical sequence quickly construct model, extract from current frame by using difference update model real-time. Then, noise caused non-background...

Precise, reliable, and low-cost vehicular localization across a continuous spatiotemporal domain is an important problem in the field of outdoor ground vehicles. This paper proposes visual odometry algorithm, where ultrarobust fast feature-matching scheme combined with effective antiblurring frame selection strategy. Our method follows procedure finding feature correspondences from consecutive frames minimizing their reprojection error. The blurred image great challenge for sharp turn or...

The accurate positioning is the core technology of mobile robot. paper proposes a visual odometry method based on trifocal tensor to get high-precision information autonomous Two-wheel car was used simulate robot, where monocular camera mounted on. We employed calibration algorithm intrinsic parameters, IPM (Inverse Perspective Mapping) top view pavement images, improved SURF detect and match feature points, calculate fundamental matrix after outliner points removing RANSAC pose from matrix....

In this paper, a hierarchical control method is proposed to design the vehicle longitudinal collision avoidance controller. After establishing inverse dynamics model, upper layer acceleration model predictive controller and underlying single neuron PID are designed respectively. The simulation results show that can successfully avoid satisfy safety, stability comfort.

This paper proposes a new method to extract background and segment targets from point clouds collected by three-dimensional roadside LiDAR in snowfall weather. Background extraction target segmentation are two main problems environmental perception based on LiDAR. first introduces filtering algorithm, which uses the historical cloud sequence construct model real time filter difference. Then, non-background algorithm is proposed, including linear density for snow noise hierarchical clustering...

It has been well recognized that driving behaviors significantly impact fuel consumption of vehicles.In this paper, we propose a FuelNet model based on Long Shortterm Memory Neural Network (LSTM NN), which can predict vehicle in very accurate manner.First, take the kinetic parameters and corresponding to build model, analyze correlations between prediction accuracy different combinations input parameters.In addition, our exhibits superior capability for (FCP) at speed, comparison with deep...

A moving horizon H/sub /spl infin// disturbance attenuation scheme for active suspensions with time-domain constraints is suggested, where closed-loop dissipation and hence performance are guaranteed. By minimizing the level online, system able to relax requirement so as meet hard in case of unforeseen large disturbances, recover a better generally mild disturbances. Simulation results 2 DOF quarter-car model presented.