Many people die each year in the world single vehicle roadway departure crashes caused by driver inattention, especially on freeway. Lane Departure Warning System (LDWS) is a useful system to avoid those accident, which, lane detection key issue. In this paper, after brief overview of existing methods, we present robust algorithm based geometrical model and Gabor filter. This two assumptions: road front approximately planar marked which are often correct highway freeway where most accidents...

The recognition and tracking of traffic lights for intelligent vehicles based on a vehicle-mounted camera are studied in this paper. candidate region the light is extracted using threshold segmentation method morphological operation. Then, algorithm machine learning employed. To avoid false negatives loss, target CAMSHIFT (Continuously Adaptive Mean Shift), which uses color histogram as model, adopted. In addition to signal pre-processing learning, initialization problem search window...

In this paper, we introduce a novel and efficient hybrid trajectory planning method for autonomous driving in highly constrained environments. The contributions of paper are fourfold. First, present framework that is able to handle geometry constraints, nonholonomic dynamics constraints cars humanlike layered fashion generate curvature-continuous, kinodynamically feasible, smooth, collision-free trajectories real time. Second, derivative-free global path modification algorithm extract...

Autonomous vehicles need to perform social accepted behaviors in complex urban scenarios including human-driven with uncertain intentions. This leads many difficult decision-making problems, such as deciding a lane change maneuver and generating policies pass through intersections. In this paper, we propose an intention-aware algorithm solve challenging problem uncontrolled intersection scenario. order consider intentions, first develop continuous hidden Markov model predict both the...

Color grading is a crucial step in the processing of fruits and vegetables that directly affects profitability, because quality agricultural products often associated with their color. Most existing automatic color systems determine either by comparing product against predefined fixed set reference colors or using separating parameters, three-dimensional spaces. Using these methods, it not convenient for user to adjust preferences parameters. In this paper, we present an effective...

Road detection is an important problem with application to driver assistance systems and autonomous, self-guided vehicles. The focus of this paper on the feature extraction classification for front-view road detection. Specifically, we propose using Support Vector Machines (SVM) effective approach self-supervised online learning. proposed algorithm capable automatically updating training data which reduces possibility misclassifying non-road classes improves adaptability algorithm. presented...

Place recognition is an important component for autonomous vehicles to achieve loop closing or global localization. In this article, we tackle the problem of place based on sequential 3-D LiDAR scans obtained by onboard sensor. We propose a transformer-based network named SeqOT exploit temporal and spatial information provided range images generated from data. It uses multiscale transformers generate descriptor each sequence in end-to-end fashion. During online operation, our finds similar...

LiDAR-based place recognition (LPR) is one of the most crucial components autonomous vehicles to identify previously visited places in GPS-denied environments. Most existing LPR methods use mundane representations input point cloud without considering different views, which may not fully exploit information from LiDAR sensors. In this article, we propose a <underline xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">c</u> ross-...

Simultaneous localization and mapping (SLAM), as an important tool for vehicle positioning mapping, plays role in the unmanned technology. This paper mainly presents a new solution to LIDAR-based SLAM vehicles off-road environment. Many methods have been proposed solve problems well. However, complex environment, especially it is difficult obtain stable results due rough road scene diversity. We propose algorithm based on grid which combining probability feature by Expectation-maximization...

Abstract As intelligent vehicles usually have complex overtaking process, a safe and efficient automated system (AOS) is vital to avoid accidents caused by wrong operation of drivers. Existing AOSs rarely consider longitudinal reactions the overtaken vehicle (OV) during overtaking. This paper proposed novel AOS based on hierarchical reinforcement learning, where reaction given data-driven social preference estimation. incorporates two modules that can function in different phases. The first...

As a core part of an autonomous driving system, motion planning plays important role in safe driving. However, traditional model- and rule-based methods lack the ability to learn interactively with environment, learning-based still have problems terms reliability. To overcome these problems, hybrid framework (HMPF) is proposed improve performance planning, which composed behavior optimization-based trajectory planning. The module adopts deep reinforcement learning (DRL) algorithm, can from...

In this paper, we present a complete, flexible and safe convex-optimization-based method to solve speed planning problems over fixed path for autonomous driving in both static dynamic environments. Our contributions are five fold. First, summarize the most common constraints raised various scenarios as requirements planner developments metrics measure capacity of existing planners roughly driving. Second, introduce more general, complete mathematical model including all summarized compared...

This paper describes the use of machine learning methods to build a decision support system for predicting distribution coastal ocean algal blooms based on remote sensing data in Monterey Bay. can help scientists obtain prior information large region and formulate strategies deploying robots more detailed situ exploration. The difficulty is that there are insufficient create direct statistical model with satellite inputs. To solve this problem, we built Random Forest using MODIS MERIS...

Two-speed transmission system is usually employed by an electric vehicle to improve efficiency of driving and drivability powered solely power, i.e. pure vehicle, or series hybrid vehicle. The improvement the due application multi-speed shown based on motor drive characteristics, traction characteristics cost. Moreover, in addition optimization automatic manual (AMT) system, integrated control technology AMT without clutch also introduced. With result conducting field test on-road test, it...

We present an iterative linear quadratic regulator (ILQR) method for trajectory tracking control of a wheeled mobile robot system. The proposed scheme involves kinematic model linearization technique, global generation algorithm, and controller design. A lattice planner, which searches over 3D (x, y, θ) configuration space, is adopted to generate the trajectory. ILQR used design local controller. effectiveness demonstrated in simulation experiment with significantly asymmetric differential...

VPH+, an enhanced vector polar histogram method (VPH), is developed for mobile robot equipped with laser radar to avoid obstacles more efficiently in complicated environment. The isolated obstacle points detected by are grouped into different blocks, so the can predict some outside its safe distance and them advance. Meanwhile, speed of heading deviation (between robot-goal direction) combined cost function which aims find out desired direction reach goal minimum time. efficiency capability...

Simultaneous Localization And Mapping (SLAM) plays a more and important role in the environment perception system of Unmanned Ground Vehicle (UGV), most SLAM technologies used to be applied indoor or urban scenarios, we present real-time 6D approach suitable for large scale natural terrain with help an Inertial Measurement Unit(IMU) two 3D Lidars. Besides dividing entire map into many submaps which consists numbers tree structure based voxels, use probabilistic methods represent possibility...

Precise understanding of the mobility is essential for high performance autonomous tracked vehicles in challenging circumstances, though complex track/terrain interaction difficult to model. A slip model based on instantaneous centers rotation (ICRs) treads presented and identified predict motion vehicle a short term. Unlike many research studies estimating current ICRs locations using velocity measurements feedback controllers, we focus predicting forward trajectories by position...

This paper proposes the decision-making framework of lane change behavior based on Hierarchical State Machine (HSM) and we build distributed control system architecture RCS (Real-Time Control System) to test model. Environment perception module, decision planning module execution are put into improve real-time ensure that several modules run simultaneously. Besides, consists two parts: miniature scene information model multi-attribute decision-making. The is HSM it sets top-level state...

Place recognition is one of the most crucial modules for autonomous vehicles to identify places that were previously visited in GPS-invalid environments.Sensor fusion considered an effective method overcome weaknesses individual sensors.In recent years, multimodal place fusing information from multiple sensors has gathered increasing attention.However, existing methods only use limited field-of-view camera images, which leads imbalance between features different modalities and limits...

This paper presents an approach to detect and recognize traffic signs present in the urban scenes China. The algorithm is composed of three steps that are color segmentation, shape detection pictogram recognition. In first step Ridge Regression used obtain a precise segmentation RGB space achieves same good performance as many machine learning based methods while using less computation time. Recognition process include novel feature extraction involves OTSU method, extracted robust against...

To reduce the data size of metric map and matching computational cost in unmanned ground vehicle self-driving navigation urban scenarios, a metric-topological hybrid system is proposed this paper. According to different positioning accuracy requirements, areas are divided into strong constraint (SC) areas, such as roads with lanes, loose (LC) intersections open areas. As direction provided by traffic lanes global waypoints road network, simple topological fit for SC While LC mainly relies on...