Smooth Behavioral Estimation For Ramp Merging Control In Autonomous Driving

A P-STABLE EIGHTEENTH-ORDER SIX-STEP METHOD FOR PERIODIC INITIAL VALUE PROBLEMS

<div class="section abstract"><div class="htmlview paragraph">Segment with lane drops are very important in freeway systems since they are major constrains to traffic flow and safety. The frequency of capacity reductions and higher safety risks is proportional to an increase in lane-changing actions, which worsen traffic congestion, decrease road capacity, and increase the risk of an accident. Traditional traffic management strategies that rely on physical structures and driver’s decision making often fail under such conditions. This paper provides a detailed lane change control strategy specific to freeway segments with lane reduction in the connected and autonomous vehicle (CAV) environment. The strategy combines both centralized and decentralized techniques to improve the vehicle’s lane-changing behavior and density. A cellular transmission model of lane-level is proposed for the centralized control of the linked vehicles based on the ratio of the driver compliance. The model derives the density equation and transforms the lane-changing problem before the work zone into merging traffic flow problem. The optimization model is developed based on the total trip time, density deviation, and total lane changes, with constraints on the cell reception capacity and lane changing ratios. Control parameters for lane change distribution are identified using genetic algorithms to solve the problem. For the decentralized control, a reinforcement learning solution is introduced which uses deep Q-networks (DQN) to improve lane-changing actions. The reward function takes into account the traffic efficiency and the impact of lane changing, and the continuous action space is discretized for application. The control mechanism is evaluated by the simulation of a work zone scenario that includes two restricted lanes on the Shanghai-Nanjing Expressway. It also shows that there is an improvement of 3% to 6% in traffic flow and velocity as compared to single-strategy approaches. The collaborative control strategy significantly enhances traffic flow and reduces congestion at bottlenecks and offers valuable information for future traffic control in CAV environments.</div></div>

Along with the rapid development of autonomous driving technology, autonomous vehicles are showing a trend of practicality and popularity. Autonomous vehicles perceive environmental information through sensors to provide a basis for the decision making of vehicles. Based on this, this paper investigates the lane-changing decision-making behavior of autonomous vehicles. First, the similarity between autonomous vehicles and moving molecules is sought based on a system-similarity analysis. The microscopic lane-changing behavior of vehicles is analyzed by the molecular-dynamics theory. Based on the objective quantification of the lane-changing intention, the interaction potential is further introduced to establish the molecular-dynamics lane-changing model. Second, the relationship between the lane-changing initial time and lane-changing completed time, and the dynamic influencing factors of the lane changing, were systematically analyzed to explore the influence of the microscopic lane-changing behavior on the macroscopic traffic flow. Finally, the SL2015 lane-changing model was compared with the molecular-dynamics lane-changing model using the SUMO platform. SUMO is an open-source and multimodal traffic experimental platform that can realize and evaluate traffic research. The results show that the speed fluctuation of autonomous vehicles under the molecular-dynamics lane-changing model was reduced by 15.45%, and the number of passed vehicles was increased by 5.93%, on average, which means that it has better safety, stability, and efficiency. The molecular-dynamics lane-changing model of autonomous vehicles takes into account the dynamic factors in the traffic scene, and it reasonably shows the characteristics of the lane-changing behavior for autonomous vehicles.

This article describes the design, implementation, and evaluation of an active lane change control algorithm for autonomous vehicles with human factor considerations. Lane changes need to be performed considering both driver acceptance and safety with surrounding vehicles. Therefore, autonomous driving systems need to be designed based on an analysis of human driving behavior. In this article, manual driving characteristics are investigated using real-world driving test data. In lane change situations, interactions with surrounding vehicles were mainly investigated. And safety indices were developed with kinematic analysis. A safety indices–based lane change decision and control algorithm has been developed. In order to improve safety, stochastic predictions of both the ego vehicle and surrounding vehicles have been conducted with consideration of sensor noise and model uncertainties. The desired driving mode is decided to cope with all lane changes on highway. To obtain desired reference and constraints, motion planning for lane changes has been designed taking stochastic prediction-based safety indices into account. A stochastic model predictive control with constraints has been adopted to determine vehicle control inputs: the steering angle and the longitudinal acceleration. The proposed active lane change algorithm has been successfully implemented on an autonomous vehicle and evaluated via real-world driving tests. Safe and comfortable lane changes in high-speed driving on highways have been demonstrated using our autonomous test vehicle.

Under manufacturing process variation, a path through a fault site is called longest for delay test if there exists a process condition under which the path has the maximum delay among all paths through that fault site. There are often multiple longest paths for each fault site in the circuit, due to different process conditions. To detect the smallest delay fault, it is necessary to test all longest paths through the fault site. However, previous methods are either inefficient or their results include too many paths that are not longest. We present an efficient method to generate the longest path set for delay test under process variation. To capture both structural and systematic process correlation, we use linear delay functions to express path delays under process variation. A novel path-pruning technique is proposed to discard paths that are not longest, resulting in a significantly reduction in the number of paths compared with the previous best method. The new method can be applied to any process variation as long as its impact on delay is linear.

3D measurement system based on divergent multi-line structured light projection, its accuracy analysis

As part of a study aimed at gaining a clearer understanding of the pathogenesis of Naegleria fowleri, we have developed a simple and reliable technique for cloning trophozoites. Although cloning of N. fowleri has been reported using limiting dilution (Jonckheere, 1979, Path. Biol. 27: 453-458) and by micropipette technique (Wong et al., 1977, J. Parasit. 63: 872-878), the degree of success is difficult to assess since no data were provided on plating efficiency for either method. We have modified a previously described procedure for cloning Naegleria gruberi (Fulton, 1970, Meth. Cell Physiol. 4: 341-475) for use with N. fowleri which has proved to be highly reproducible and provides excellent efficiency of plating. Naegleria fowleri Lee (ATCC-30894) is routinely maintained in Nelson medium containing 2% calf serum as previously described (Weik and John, 1977, J. Parasit. 63: 868-871). Amebae used in cloning experiments were grown in tissue culture flasks (75 cm2, Falcon Plastics, Oxnard, California) containing 50 ml of Nelson medium supplemented with 2% calf serum. Flasks are inoculated with 104 amebae/ml and allowed to incubate for 72 hr at 37 C. The amebae were dislodged by placing the flasks on ice for 10 min and the resulting suspension of cells was vortexed to dissociate any remaining clumps of organisms. After quantitation by hemacytometer count and in some experiments by Coulter counter (Weik and John, 1977, J. Parasit. 63:150-151) the amebae were diluted to a final concentration of 1030 amebae/ml in Nelson medium without calf serum. The Escherichia coli (hospital isolate) used in cloning N. fowleri were first grown for 3 hr at 37 C with agitation in 50 ml of Nutrient Broth (Difco). The bacteria were diluted 1/10, 1/ 100, and 1/1,000 in soft nutrient agar (Difco, 0.5% agar w/v) and maintained at 55 C in a heated waterbath. The dilution was made in the following manner: 9.0 ml of bacteria in soft nutrient agar are placed in the heated waterbath and then 1.0 ml of Nelson medium containing the diluted amebae was added. The tube is then immediately vortexed and poured onto petri plates containing a non-nutrient agar base (1.5% agar, 10 ml per plate). Control plates containing no amebae were included with each experiment to assure that the observed plaques were not due to bacteriophage. The soft agar plates were allowed to harden at room temperature for 5 hr and then placed upright in a 37 C incubator stacked in sets of 3. After 24 hr incubation, parafilm was placed around the petri plates to prevent drying of the medium. The results from a representative cloning exp riment are shown in Table I. Plaques containing amebae are regularly observed only after 5 to 6 days of incubation with the presence of amebae being confirmed by phase-contrast microscopy. It is especially important to inspect plates carefully after this period since plaques will merge or disappear within the succeeding 24-48 hr. Amebae in the plaque can be observed growing in a concentric manner, a characteristic feature of Naegleria that has been previously described (Carter, 1970, J. Pathol. 100: 217-244). For amplification of clones, amebal plaques are aseptically cut out of the agar plates and placed in Falcon flasks containing Nelson medium supplemented with 2% calf serum and 100 ,tg/ml streptomycin. Two of the amebal clones tested so far have maintained their infectivity for mice. A notable feature of these experiments is the high efficiency of plating (100%) obtained when 3,000 bacteria/ml are initially diluted into the overlay whereas no plaques are observed when 300 or 30 bacteria/ml are used. We have had success using bacterial concentrations between 3,000 and 9,000 bacteria/ml in the initial overlay. The cloning method described in this report differs from previous methods in the use of the soft agar overlay. The observed high efficiency of plating may be due in part to the more effective separation of the individual amebae during the plating procedure. A previous cloning method for N. gruberi employed a glass spreading bar to

Landform recognition is one of the most significant aspects of geomorphology research, which is the essential tool for landform classification and understanding geomorphological processes. Watershed object-based landform recognition is a new spot in the field of landform recognition. However, in the relevant studies, the quantitative description of the watershed generally focused on the overall terrain features of the watershed, which ignored the spatial structure and topological relationship, and internal mechanism of the watershed. For the first time, we proposed an effective landform recognition method from the perspective of the watershed spatial structure, which is separated from the previous studies that invariably used terrain indices or texture derivatives. The slope spectrum method was used herein to solve the uncertainty issue of the determination on the watershed area. Complex network and P–N terrain, which are two effective methodologies to describe the spatial structure and topological relationship of the watershed, were adopted to simulate the spatial structure of the watershed. Then, 13 quantitative indices were, respectively, derived from two kinds of watershed spatial structures. With an advanced machine learning algorithm (LightGBM), experiment results showed that the proposed method showed good comprehensive performances. The overall accuracy achieved 91.67% and the Kappa coefficient achieved 0.90. By comparing with the landform recognition using terrain indices or texture derivatives, it showed better performance and robustness. It was noted that, in terms of loess ridge and loess hill, the proposed method can achieve higher accuracy, which may indicate that the proposed method is more effective than the previous methods in alleviating the confusion of the landforms whose morphologies are complex and similar. In addition, the LightGBM is more suitable for the proposed method, since the comprehensive manifestation of their combination is better than other machine learning methods by contrast. Overall, the proposed method is out of the previous landform recognition method and provided new insights for the field of landform recognition; experiments show the new method is an effective and valuable landform recognition method with great potential as well as being more suitable for watershed object-based landform recognition.

Autonomous driving has attracted significant research interests in the past two decades as it offers many potential benefits, including releasing drivers from exhausting driving and mitigating traffic congestion, among others. Despite promising progress, lane-changing remains a great challenge for autonomous vehicles (AV), especially in mixed and dynamic traffic scenarios. Recently, reinforcement learning (RL) has been widely explored for lane-changing decision makings in AVs with encouraging results demonstrated. However, the majority of those studies are focused on a single-vehicle setting, and lane-changing in the context of multiple AVs coexisting with human-driven vehicles (HDVs) have received scarce attention. In this paper, we formulate the lane-changing decision-making of multiple AVs in a mixed-traffic highway environment as a multi-agent reinforcement learning (MARL) problem, where each AV makes lane-changing decisions based on the motions of both neighboring AVs and HDVs. Specifically, a multi-agent advantage actor-critic (MA2C) method is proposed with a novel local reward design and a parameter sharing scheme. In particular, a multi-objective reward function is designed to incorporate fuel efficiency, driving comfort, and the safety of autonomous driving. A comprehensive experimental study is made that our proposed MARL framework consistently outperforms several state-of-the-art benchmarks in terms of efficiency, safety, and driver comfort.

The development of automated driving vehicles aims to provide safer, comfortable, and more efficient mobility options. However, the decision-making control of autonomous vehicles still faces limitations of human performance mimicry. These limitations become particularly evident in complex and unfamiliar driving scenarios, where weak decision-making abilities and poor adaptation of vehicle behaviour are prominent issues. This paper proposes a game-theoretic decision-making algorithm for human-like driving in the vehicle lane change scenario. Firstly, an inverse reinforcement learning (IRL) model is used to quantitatively analyze the lane change trajectories of the natural driving dataset, establishing the human-like human cost function. Subsequently, joint safety, and comfort to build the comprehensive decision cost function. The combined decision cost function is used to conduct a noncooperative game of vehicle lane changing decisions to solve the optimal decision of host vehicle lane changing. The host vehicle lane-changing decision problem is formulated as a Stackelberg game optimization problem. To verify the feasibility and effectiveness of the algorithm proposed in this study, a lane change test scenario was established. Firstly, we analyze the human-like decision-making model derived from the maximum entropy inverse reinforcement learning algorithm to verify the effectiveness and robustness of the IRL algorithm. Secondly, the human-like game decision-making algorithm in this paper is validated by conducting an interactive lane-changing experiment with obstacle vehicles of different driving styles. The experimental results prove that the human-like driving decision-making model proposed in this study can make lane-changing behaviours in line with human driving patterns in lane-changing scenarios of the expressway.

Purpose The purpose of this paper is to investigate problems in performing stable lane changes and to find a solution to reduce energy consumption of autonomous electric vehicles. Design/methodology/approach An optimization algorithm, model predictive control (MPC) and Karush–Kuhn–Tucker (KKT) conditions are adopted to resolve the problems of obtaining optimal lane time, tracking dynamic reference and energy-efficient allocation. In this paper, the dynamic constraints of vehicles during lane change are first established based on the longitudinal and lateral force coupling characteristics and the nominal reference trajectory. Then, by optimizing the lane change time, the yaw rate and lateral acceleration that connect with the lane change time are limed. Furthermore, to assure the dynamic properties of autonomous vehicles, the real system inputs under the restraints are obtained by using the MPC method. Based on the gained inputs and the efficient map of brushless direct-current in-wheel motors (BLDC IWMs), the nonlinear cost function which combines vehicle dynamic and energy consumption is given and the KKT-based method is adopted. Findings The effectiveness of the proposed control system is verified by numerical simulations. Consequently, the proposed control system can successfully achieve stable trajectory planning, which means that the yaw rate and longitudinal and lateral acceleration of vehicle are within stability boundaries, which accomplishes accurate tracking control and decreases obvious energy consumption. Originality/value This paper proposes a solution to simultaneously satisfy stable lane change maneuvering and reduction of energy consumption for autonomous electric vehicles. Different from previous path planning researches in which only the geometric constraints are involved, this paper considers vehicle dynamics, and stability boundaries are established in path planning to ensure the feasibility of the generated reference path.

With the fast development of autonomous vehicle (AV) technology and possible popularity of AVs in the near future, a mixed-vehicle type driving environment where both AVs and their surrounding human-driving vehicles drive on the same road will exist and last for a long time. An AV measures its driving environments in real time and make control decisions to ensure driving safety. However, surrounding human-driving vehicles may conduct aggressive driving behaviors (e.g., sudden deceleration, sudden acceleration, sudden left or right lane change) in practice, which requires an AV to make correct control decisions to eliminate the effect of aggressive driving behaviors on its driving safety. In this paper, we propose a reinforcement learning based decision-making system (ReDS) which considers aggressive driving behaviors of surrounding human-driving vehicles during the decision making process. In ReDS, we firstly build a mixture density network based aggressive driving behavior detection method to detect possible aggressive driving behaviors among surrounding vehicles of an AV. We then build a reward function based on aggressive driving behavior detection results and incorporate the reward function into a reinforcement learning model to make optimal control decisions considering aggressive driving behaviors. We use a real-world traffic dataset from the United States Department of Transportation Federal Highway Administration to evaluate optimal control decision determination performance of ReDS in comparison with the state-of-the-art methods. The comparison results show that ReDS can improve optimal control decision success rate by 43% compared with existing methods, which demonstrates that ReDS has good optimal control decision determination performance.

A dynamic lane-changing decision and trajectory planning model of autonomous vehicles under mixed autonomous vehicle and human-driven vehicle environment

According to the previous research, lane-changes are a major cause of serious traffic accidents. Thus, it is essential to build an efficient prediction algorithm for vehicles lane change on Advanced Driving Assistance System (ADAS) to ensure a safe and comfort driving for host vehicle. At present, many methods for lane change prediction have been proposed. However, most of them require a lot of data training to have a good prediction performance. Considering the practical applicability of the prediction algorithm in ADAS, this paper proposes a prediction method by probabilistic modelling. This method combines two aspects lane change evidence. On the one hand, model the lane change probability caused by the target vehicle's driving context. On the other hand, model the lane change probability reflected by vehicle posture in the road. The evaluation of the whole algorithms was done by using simulation data and real lane change data. The results show that the algorithm performs well in predicting accuracy and reducing false alarms.

Many computer vision applications can be formulated as labeling problems. However, multilabeling problems are usually very challenging to solve, especially when some ordering constraints are enforced. We solve in this paper a five-parts labeling problem proposed in [6, 7]. In this model, one wants to find an optimal labeling for an image with five possible parts: “left”, “right”, “top”, “bottom” and “center”. The geometric ordering constraints can be read naturally from the names. No previous method can solve the problem with globally optimal solutions in a linear space complexity. We propose an efficient dynamic programming based algorithm which guarantees the global optimal labeling for the five-parts model. The time complexity is O(N1.5) and the space complexity is O(N), with N being the number of pixels in the image. In practice, it runs faster than previous methods. Moreover, it works for both 4-neighborhood and 8-neighborhood settings, and can be easily parallelized for GPU.