02169nas a2200265 4500008004100000245008000041210006900121260001200190300001300202520135300215653002801568653002101596100001901617700002201636700002201658700001801680700001901698700001801717700003101735700002201766700002301788700002301811700002101834856004801855 2020 eng d00aAre commercially implemented adaptive cruise control systems string stable?0 aAre commercially implemented adaptive cruise control systems str c06/2020 a12 pages3 a
In this article, we assess the string stability of seven 2018 model year adaptive cruise control (ACC) equipped vehicles that are widely available in the US market. A total of seven distinct vehicle models from two different vehicle makes are analyzed using data collected from more than 1,200 miles of driving in designed car-following experiments with ACC engaged by the following vehicle. The data is used to identify the parameters of a linear second order delay differential equation model that approximates the behavior of the proprietary ACC systems. The string stability of the data fitted model associated with each vehicle is assessed, and the main finding is that all seven vehicle models have string unstable ACC systems. For one commonly available vehicle that offers ACC as a standard feature on all trim levels, we validate the string stability finding with a multi-vehicle platoon experiment in which all vehicles are the same year, make, and model. In the multi-vehicle platoon test, an initial disturbance of 6 mph is amplified by 19 mph to a 25 mph disturbance, at which point the last vehicle in the platoon is observed to disengage the ACC and return control to the human driver. The data collected the driving experiments is made available, representing the largest available driving dataset on ACC equipped vehicles.
10aAdaptive Cruise Control10aString Stability1 aGunter, George1 aGloudemans, Derek1 aStern, Raphael, E1 aMcQuade, Sean1 aBhadani, Rahul1 aBunting, Matt1 aMonache, Maria, Laura Dell1 aSeibold, Benjamin1 aSprinkle, Jonathan1 aPiccoli, Benedetto1 aWork, Daniel, B. uhttp://dx.doi.org/10.1109/TITS.2020.300068201575nas a2200193 4500008004100000245009200041210006900133520087400202653002501076653002101101653002901122653001501151653002601166100001601192700001801208700001901226700002301245856011301268 2020 eng d00aModeling Human Car-Following Behavior from Demonstration with Recurrent Neural Networks0 aModeling Human CarFollowing Behavior from Demonstration with Rec3 aThe validity of simulation testing for autonomous vehicles depends on the ability to accurately simulate human driving behavior. This project seeks to train a model on an individual’s driving data, and to test the ability of the model to predict trajectories that replicate the driver’s style by using the model in a realistic simulated environment. Specifically, we deployed Recurrent Neural Network (RNN) modeling techniques to create a black-box model of an individual’s driving behavior. We use our RNN-trained model to simulate a human-driven vehicle in the Robot Operating System (ROS) based CAT Vehicle simulator for autonomous vehicle validation. We hope this work is a step to improve testing environments for validating human behavior replicating car-following models and thereby improve testing environments for autonomous vehicles in general.
10acar-following models10adriving behavior10arecurrent neural network10asimulation10atrajectory prediction1 aJones, Iris1 aWalter, Megan1 aBhadani, Rahul1 aSprinkle, Jonathan uhttp://csl.arizona.edu/content/modeling-human-car-following-behavior-demonstration-recurrent-neural-networks01020nas a2200205 4500008004100000020002200041245010800063210006900171260005200240520028500292653001400577653003300591653001900624100001900643700001900662700002500681700002300706700001700729856006800746 2020 eng d a978-1-943580-80-400aProgramming the Kennedy Receiver for Capacity Maximization versus Minimizing One-shot Error Probability0 aProgramming the Kennedy Receiver for Capacity Maximization versu aOnlinebThe Optical Society of Americac09/20203 aWe find the capacity attained by the Kennedy receiver for coherent-state BPSK when the symbol prior $p$ and pre-detection displacement $\beta$ are optimized. The optimal $\beta$ is different than what minimizes error probability for single-shot BPSK state discrimination.
10aPhotonics10aPhotonics Information Theory10aQuantum Optics1 aBhadani, Rahul1 aGrace, Michael1 aDjordjevic, Ivan, B.1 aSprinkle, Jonathan1 aGuha, Saikat uhttps://www.osapublishing.org/abstract.cfm?uri=FiO-2020-JM6B.2900570nas a2200133 4500008004100000245011400041210006900155260000700224100002200231700002100253700001900274700002300293856012000316 2020 eng d00aSafety and Stability Analysis of the FollowerStopper Traffic Wave Dampening Controller (Late-Breaking Poster)0 aSafety and Stability Analysis of the FollowerStopper Traffic Wav c071 aKreienkamp, Chris1 aFishbein, Daniel1 aBhadani, Rahul1 aSprinkle, Jonathan uhttp://csl.arizona.edu/content/safety-and-stability-analysis-followerstopper-traffic-wave-dampening-controller-late00608nas a2200181 4500008004100000245010100041210006900142260001200211653002400223653002000247653002100267100001800288700002000306700001800326700001900344700001600363856004700379 2019 eng d00aAdaptive HSL Filters and Inverse Perspective Transforms in Lane Detection for Autonomous Driving0 aAdaptive HSL Filters and Inverse Perspective Transforms in Lane c03/201910aautonomous vehicles10acomputer vision10aimage processing1 aMason, Hannah1 aBentley, Landon1 aMacInnes, Joe1 aBhadani, Rahul1 aBose, Tamal uhttp://doi.org/10.13140/RG.2.2.18919.6032701697nas a2200181 4500008004100000245006200041210006200103260004700165520109000212653002401302653000901326653001501335100002001350700001601370700001901386700002101405856008901426 2019 eng d00aDomain Specific Modeling Language for Test World Creation0 aDomain Specific Modeling Language for Test World Creation aTucsonbThe University of Arizonac08/20193 aIt is often necessary to use a 3D physics simulator in order to model and test complex robotic systems. Verifying certain behaviors of systems in the real world can be costly, time-consuming, and even dangerous, while simulations are relatively cheap and fast. However, creating simulated environments to test robotic behaviors can take up quite a lot of time and processing power. In order for behaviors to be tested in a variety of scenarios, multiple environments must be created, causing verification time to increase. This paper presents a domain-specific modeling language that can be used to speed up this process. This modeling language can be used in WebGME to generate multiple world and launch files in Gazebo, a 3D dynamics simulator. These world files can then be used to test various behaviors of complex robots such as the CAT Vehicle (Cognitive and Autonomous Test Vehicle) in a variety of simulated environments. This model language can save valuable testing time by quickly creating usable test files for complex physics based models such as the CAT Vehicle.
10aautonomous vehicles10adsml10asimulation1 aAlexander, Jill1 aPyryt, Alex1 aBhadani, Rahul1 aBunting, Matthew uhttp://csl.arizona.edu/content/domain-specific-modeling-language-test-world-creation02086nas a2200265 4500008004100000020002200041245009400063210006900157260004400226300001400270520121200284100003101496700002001527700001601547700002201563700001901585700002201604700002301626700002101649700002301670700002701693700002401720700002701744856004901771 2019 eng d a978-3-030-25446-900aFeedback Control Algorithms for the Dissipation of Traffic Waves with Autonomous Vehicles0 aFeedback Control Algorithms for the Dissipation of Traffic Waves aChambSpringer International Publishing a275–2993 aThis article considers the problem of traffic control in which an autonomous vehicle is used to regulate human-piloted traffic to dissipate stop-and-go traffic waves. We first investigated the controllability of well-known microscopic traffic flow models, namely, (i) the Bando model (also known as the optimal velocity model), (ii) the follow-the-leader model, and (iii) a combined optimal velocity follow-the-leader model. Based on the controllability results, we proposed three control strategies for an autonomous vehicle to stabilize the other, human-piloted traffics. We subsequently simulate the control effects on the microscopic models of human drivers in numerical experiments to quantify the potential benefits of the controllers. Based on the simulations, finally, we conduct a field experiment with 22 human drivers and a fully autonomous-capable vehicle, to assess the feasibility of autonomous vehicle-based traffic control on real human-piloted traffic. We show that both in simulation and in the field test that an autonomous vehicle is able to dampen waves generated by 22 cars, and that as a consequence, the total fuel consumption of all vehicles is reduced by up to 20{%}.
1 aMonache, Maria, Laura Dell1 aLiard, Thibault1 aRat, Anaïs1 aStern, Raphael, E1 aBhadani, Rahul1 aSeibold, Benjamin1 aSprinkle, Jonathan1 aWork, Daniel, B.1 aPiccoli, Benedetto1 aBlondin, Maude, Josée1 aPardalos, Panos, M.1 aSáez, Javier, Sanchis uhttps://doi.org/10.1007/978-3-030-25446-9_1202406nas a2200157 4500008004100000020002200041245005700063210005600120260001000176300001200186520191000198100001902108700001802127700002302145856008002168 2019 eng d a978-1-119-55239-000aModel-based engineering with application to autonomy0 aModelbased engineering with application to autonomy bWiley a255-2853 aIn this chapter we focus on where models fit into the verification and validation design cycle of autonomous cyber-physical systems. These systems typically make decisions through myriad of sensing loops, have implementations in multiple languages, and may have their logic represented in several different kinds of formal models. The use of code generation, along with software-in-the-loop and hardware-in-the-loop simulation (discussed further in Section 4), permits system designers to apply various agile techniques for the validation and verification of systems as requirements are implemented, tested, and demonstrated. The work in this chapter explores such a design cycle with application to autonomous driving. Examples are given for the implementation of various components that describe vehicle dynamics, control models, system identification, sensor/data acquisition, etc., which can be functionally de- scribed in models, and explored in simulation before utilizing code generation to deploy final solutions. The integration of simulation tools during functional design, software-in-the-loop testing, and hardware-in-the-loop testing, permits regression evaluation of use case scenarios. In addition to functional testing, we also describe how high-level domain- specific models can be used to include verification-in-the-loop toolboxes as part of the design cycle. All the examples in this chapter are based on an autonomous Ford Escape, which has a Robotic Operating System (ROS) API for its control and the integration of autonomous components—however, the results are applicable to other event-based and time-triggered middleware platforms. The implementation models in use include Simulink, MATLAB, StateFlow, and other domain-specific languages that specify high-level behaviors.
1 aBhadani, Rahul1 aBunting, Matt1 aSprinkle, Jonathan uhttp://csl.arizona.edu/content/model-based-engineering-application-autonomy02029nas a2200169 4500008004100000245009800041210006900139260011900208520129400327653002401621653002101645100002001666700001801686700001901704700001601723856012001739 2019 eng d00aA Pseudo-Derivative Method for Sliding Window Path Mapping in Robotics-Based Image Processing0 aPseudoDerivative Method for Sliding Window Path Mapping in Robot aAccepted at IEEE International Conference on Robotic ComputingbCAT Vehicle Research Experience for Undergraduates3 aA sliding window technique in robotics-based image processing applications is a common approach to path mapping from extracted features. Mapping a path inside an image requires finding a series of points representing the path. Previous approaches find these points by sliding a window along the path in fixed increments across one image dimension. After each slide, the center of the window in the other dimension is adjusted so that the window maximally covers the path in that area. This approach, however, fails to map paths that experience sharp curvature since the windows slide along only one dimension. The method proposed herein uses a pseudo-derivative approach to sliding windows that improves upon the traditional technique by dynamically adjusting the windows along both image dimensions during each slide. In this method, the directional components of a vector representing the previous slide are used as a naive estimation to perform the current slide. If this fails to map the path, the vector direction is used to enlarge the window dimensions. The method was tested in the domain of autonomous vehicles as an approach for detecting road lane markings. The algorithm proved more successful than previous sliding window approaches on perspective mapped lane images.
10aautonomous vehicles10aimage processing1 aBentley, Landon1 aMacInnes, Joe1 aBhadani, Rahul1 aBose, Tamal uhttp://csl.arizona.edu/content/pseudo-derivative-method-sliding-window-path-mapping-robotics-based-image-processing02015nas a2200253 4500008004100000245009900041210006900140250000700209260003000216520122200246653002401468653001901492653001501511653001201526100001901538700002101557700002201578700002201600700001501622700002301637700002301660700002101683856005701704 2019 eng d00aReal-Time Distance Estimation and Filtering of Vehicle Headways for Smoothing of Traffic Waves0 aRealTime Distance Estimation and Filtering of Vehicle Headways f a10 aMontreal, Canadac04/20193 aIn this paper, we describe an experience report and field deployment of real-time filtering algorithms used with a robotic vehicle to smooth emergent traffic waves. When smoothing these waves in simulation, a common approach is to implement controllers that utilize headway, relative velocity and even acceleration from smooth ground truth information, rather than from realistic data. As a result, many results may be limited in their impact when considering the dynamics of the vehicle under control and the discretized nature of the laser data as well as its periodic arrival. Our approach discusses trade-offs in estimation accuracy to provide both distance and velocity estimates, with ground-truth hardware-in-the-loop tests with a robotic car. The contribution of the work enabled an experiment with 21 vehicles, including the robotic car closing the loop at up to 8.0 m/s with this filtered estimate, stressing the importance of an algorithm that can deliver real-time results with acceptable accuracy for the safety of the drivers in the experiment.
10aautonomous vehicles10aDigital Filter10asimulation10aTraffic1 aBhadani, Rahul1 aBunting, Matthew1 aSeibold, Benjamin1 aStern, Raphael, E1 aCui, Shumo1 aSprinkle, Jonathan1 aPiccoli, Benedetto1 aWork, Daniel, B. uhttps://dl.acm.org/citation.cfm?doid=3302509.331402601205nas a2200181 4500008004100000245007600041210006900117260004700186520056100233653002300794653002100817100001700838700001700855700001300872700001900885700001600904856010300920 2019 eng d00aReinforcement Learning for Autonomous Driving using CAT Vehicle Testbed0 aReinforcement Learning for Autonomous Driving using CAT Vehicle aTucsonbThe University of Arizonac08/20193 a
We discuss a deep reinforcement learning implementation using the CAT Vehicle Testbed and discuss the merits of simulation-based deep reinforcement learning. After implementing a simple 3 layered neural network which learned using deep Q-learning, we found some challenges associated with ROS and Gazebo. In spite of these challenges, we demonstrate that our reinforcement learning architecture can teach a car to avoid obstacles. With these preliminary results, we discuss what new things we can do and how we can implement more advanced methods.
10aAutonomous Systems10aMachine Learning1 aNguyen, John1 aHuynh, Hoang1 aAv, Eric1 aBhadani, Rahul1 aBose, Tamal uhttp://csl.arizona.edu/content/reinforcement-learning-autonomous-driving-using-cat-vehicle-testbed01438nas a2200181 4500008004100000245005300041210005300094260004700147520082100194653002401015653002001039653003101059100002201090700002101112700001901133700002301152856008101175 2019 eng d00aSafety and stability analysis of FollowerStopper0 aSafety and stability analysis of FollowerStopper aTucsonbThe University of Arizonac08/20193 aIn this paper, we demonstrate that the velocity controller, FollowerStopper, is safe and string unstable. FollowerStopper is a controller that is meant to be implemented on an autonomous vehicle or in an adaptive cruise control (ACC) system. Through mathematical proof, simulation in Simulink, and hardware in the loop implementation on a real autonomous vehicle through the Robot Operating System (ROS) and Gazebo, several results are achieved. It is found that an autonomous vehicle controlled by FollowerStopper will never crash. FollowerStopper will dissipate larger traffic waves from human-driven vehicles but will amplify smaller velocity perturbations that are created within the controller. Given the maximum LiDAR range of 81 m, FollowerStopper will never command a velocity greater than 13.69 m/s
10aautonomous vehicles10acontrol systems10aintelligent transportation1 aKreienkamp, Chris1 aFishbein, Daniel1 aBhadani, Rahul1 aSprinkle, Jonathan uhttp://csl.arizona.edu/content/safety-and-stability-analysis-followerstopper00634nas a2200157 4500008004100000245009900041210006900140260003600209653002400245100002000269700001800289700001800307700001900325700001600344856011600360 2019 eng d00aA Sliding Window for Path Mapping based on a Pseudo-Derivative Method in Autonomous Navigation0 aSliding Window for Path Mapping based on a PseudoDerivative Meth aHonolulu, HawaiibIEEEc11/201910aautonomous vehicles1 aBentley, Landon1 aMacInnes, Joe1 aMason, Hannah1 aBhadani, Rahul1 aBose, Tamal uhttp://csl.arizona.edu/content/sliding-window-path-mapping-based-pseudo-derivative-method-autonomous-navigation00795nas a2200253 4500008004100000245009500041210006900136260001300205300001300218490000700231100001500238700002200253700001500275700003100290700001900321700002100340700002100361700002400382700002300406700002200429700002300451700002100474856004600495 2019 eng d00aTracking vehicle trajectories and fuel rates in phantom traffic jams: Methodology and data0 aTracking vehicle trajectories and fuel rates in phantom traffic bElsevier a82–1090 v991 aWu, Fangyu1 aStern, Raphael, E1 aCui, Shumo1 aMonache, Maria, Laura Dell1 aBhadani, Rahul1 aBunting, Matthew1 aChurchill, Miles1 aHamilton, Nathaniel1 aPiccoli, Benedetto1 aSeibold, Benjamin1 aSprinkle, Jonathan1 aWork, Daniel, B. uhttps://doi.org/10.1016/j.trc.2018.12.01201253nas a2200241 4500008004100000245008200041210006900123520049300192653002800685653002100713100001900734700001300753700002200766700002100788700003100809700001900840700001800859700001900877700002300896700002200919700002300941856004700964 2019 eng d00aWiP Abstract: String stability of commercial adaptive cruise control vehicles0 aWiP Abstract String stability of commercial adaptive cruise cont3 aIn this work, we conduct a series of car-following experiments with seven different ACC vehicles and use the collected data to model the car-following behavior of each vehicle. Using a linear stability analysis, the string stability of each tested vehicle is analyzed. Addition- ally, platoon experiments with platoons of up to eight identical vehicles are conducted to validate the stability findings. Previously, only one commercial ACC system has been evaluated for string stability. 10aAdaptive Cruise Control10aString Stability1 aGunter, George1 aYang, Y.1 aStern, Raphael, E1 aWork, Daniel, B.1 aMonache, Maria, Laura Dell1 aBhadani, Rahul1 aBunting, Matt1 aLysecky, Roman1 aSprinkle, Jonathan1 aSeibold, Benjamin1 aPiccoli, Benedetto uhttps://dl.acm.org/citation.cfm?id=331332500669nas a2200229 4500008004100000245004800041210004200089100001500131700002200146700001500168700003100183700001900214700002100233700002100254700002400275700001500299700002300314700002200337700002300359700002100382856003600403 2018 eng d00aThe Arizona Ring Experiments Dataset (ARED)0 aArizona Ring Experiments Dataset ARED1 aWu, Fangyu1 aStern, Raphael, E1 aCui, Shumo1 aMonache, Maria, Laura Dell1 aBhadani, Rahul1 aBunting, Matthew1 aChurchill, Miles1 aHamilton, Nathaniel1 aWu, Fangyu1 aPiccoli, Benedetto1 aSeibold, Benjamin1 aSprinkle, Jonathan1 aWork, Daniel, B. uhttp://hdl.handle.net/1803/935800630nas a2200169 4500008004100000245010500041210006900146260001200215490000800227653002400235653001500259653001200274100001900286700002300305700002100328856011100349 2018 eng d00a{The CAT Vehicle Testbed: A Simulator with Hardware in the Loop for Autonomous Vehicle Applications}0 aCAT Vehicle Testbed A Simulator with Hardware in the Loop for Au c04/20180 v26910aautonomous vehicles10asimulation10atestbed1 aBhadani, Rahul1 aSprinkle, Jonathan1 aBunting, Matthew uhttp://csl.arizona.edu/content/cat-vehicle-testbed-simulator-hardware-loop-autonomous-vehicle-applications02161nas a2200205 4500008004100000245009600041210006900137260004100206490000700247520150000254653002401754653000801778653001201786100001901798700002301817700002201840700002301862700002101885856004901906 2018 eng d00aDissipation of Emergent Traffic Waves in Stop-and-Go Traffic Using a Supervisory Controller0 aDissipation of Emergent Traffic Waves in StopandGo Traffic Using aFontainbleau, Miami Beach, USAbIEEE0 v573 aThis paper presents the use of a quadratic band controller in an autonomous vehicle (AV) to regulate emergent traffic waves resulting from traffic congestion. The controller dampens the emergent traffic waves through modulating its velocity according to the relative distance and velocity of the immediately preceding vehicle in the flow. At the same time, it prevents any collision within the range specified by the design parameters. The approach is based on a configurable quadratic band that allows smooth transitions between (i) no modification to the desired velocity; (ii) braking to match the speed of the preceding vehicle; and (iii) braking to avoid collision with the lead vehicle. By assuming that the lead vehicle's velocity will be oscillatory, the controller's smooth transition between modes permits any vehicle following the AV to have a smoother reference velocity. The configurable quadratic band allows design parameters, such as actuator and computation delays as well as the dynamics of vehicle deceleration, to be taken into account when constructing the controller. Experimental data, software-in-the-loop distributed simulation, and results from physical platform performance in an experiment with 21 human-driven vehicles are presented. Analysis shows that the design parameters used in constructing the quadratic band controller are met, and assumptions regarding the oscillatory nature of emergent traffic waves are valid.
10aautonomous vehicles10aCPS10aTraffic1 aBhadani, Rahul1 aPiccoli, Benedetto1 aSeibold, Benjamin1 aSprinkle, Jonathan1 aWork, Daniel, B. uhttps://ieeexplore.ieee.org/document/861970000875nas a2200277 4500008004100000245009100041210006900132260001200201490000700213653002400220653002700244100002200271700001500293700003100308700001900339700002100358700002100379700002400400700002100424700001500445700002300460700002200483700002300505700002100528856004800549 2018 eng d00aDissipation of stop-and-go waves via control of autonomous vehicles: Field experiments0 aDissipation of stopandgo waves via control of autonomous vehicle c04/20180 v8910aautonomous vehicles10acyber physical systems1 aStern, Raphael, E1 aCui, Shumo1 aMonache, Maria, Laura Dell1 aBhadani, Rahul1 aBunting, Matthew1 aChurchill, Miles1 aHamilton, Nathaniel1 aPohlmann, Hannah1 aWu, Fangyu1 aPiccoli, Benedetto1 aSeibold, Benjamin1 aSprinkle, Jonathan1 aWork, Daniel, B. uhttp://csl.arizona.edu/stern2017dissipation01583nas a2200241 4500008004100000245006200041210005800103260003400161520086600195653002401061653000701085653000801092653001001100653002201110653001101132653000801143100002001151700002201171700001901193700002101212700002301233856008501256 2018 eng d00a{A LiDAR Error Model for Cooperative Driving Simulations}0 aLiDAR Error Model for Cooperative Driving Simulations aTaipei, TaiwanbIEEEc12/20183 aCooperative driving and vehicular network simulations have done huge steps toward high realism. They have become essential tools for performance evaluation of any kind of vehicular networking application. Yet, cooperative vehicular applications will not be built on top of wireless networking alone, but rather fusing together different data sources including sensors like radars, LiDARs, or cameras. So far, these sensors have been assumed to be ideal, i.e., without any measurement error. This paper analyzes a set of estimated distance traces obtained with a LiDAR sensor and develops a stochastic error model that can be used in cooperative driving simulations. After implementing the model within the PLEXE simulation framework, we show the impact of the model on a set of cooperative driving control algorithms.
10aautonomous vehicles10aAV10aCAV10aLiDAR10aself-driving cars10asensor10aVNC1 aSegata, Michele1 aCigno, Renato, Lo1 aBhadani, Rahul1 aBunting, Matthew1 aSprinkle, Jonathan uhttp://csl.arizona.edu/content/lidar-error-model-cooperative-driving-simulations00954nas a2200301 4500008004100000020002200041245010200063210006900165260002700234300001000261653002400271653001700295100002100312700002200333700001500355700002100370700001500391700002100406700002200427700002300449700001900472700002100491700002300512700003100535700002400566700001500590856004700605 2017 eng d a978-1-4503-4976-500aControlling for Unsafe Events in Dense Traffic Through Autonomous Vehicles: Invited Talk Abstract0 aControlling for Unsafe Events in Dense Traffic Through Autonomou aNew York, NY, USAbACM a7–710aSugiyama experiment10aTraffic flow1 aWork, Daniel, B.1 aStern, Raphael, E1 aWu, Fangyu1 aChurchill, Miles1 aCui, Shumo1 aPohlmann, Hannah1 aSeibold, Benjamin1 aPiccoli, Benedetto1 aBhadani, Rahul1 aBunting, Matthew1 aSprinkle, Jonathan1 aMonache, Maria, Laura Dell1 aHamilton, Nathaniel1 aHaulcy, R. uhttp://doi.acm.org/10.1145/3055378.305538001800nas a2200229 4500008004100000245006000041210005800101260009100159520099000250653002301240653001901263100001901282700002401301700001901325700002301344700002101367700002001388700002301408700003101431700002201462856008601484 2017 eng d00aA Fuzzy based approach to Dampen Emergent Traffic Waves0 aFuzzy based approach to Dampen Emergent Traffic Waves aThe University of ArizonabCAT Vehicle Research Experience for Undergraduatesc01/20173 aAdaptive Cruise Control (ACC) and Traffic Aware Cruise Control (TACC) are recent advancements in cruise control design that allow a semi-autonomous vehicle to slow itself when approaching vehicles. The issue with these technologies is that they focus on keeping the distance from a leading vehicle constant. This may lead to unwanted dynamics in the following traffic flow, could result in the creation of traveling waves. This paper focuses on maintaining a reference velocity based on the relative position of the preceding vehicle instead of slowing down to maintain a certain following distance. Doing so could reduce the amount of braking the vehicles behind the autonomous vehicle will do. With this kind of technology implemented, the number and duration of traffic jams could be greatly reduced. Simulation results and tests run on the University of Arizona's Cognitive Autonomous Test (CAT) Vehicle illustrate the feasibility and success of this new controller.
10aAutonomous Systems10aControl System1 aHaulcy, R'mani1 aHamilton, Nathaniel1 aBhadani, Rahul1 aSprinkle, Jonathan1 aWork, Daniel, B.1 aRisso, Nathalie1 aPiccoli, Benedetto1 aMonache, Maria, Laura Dell1 aSeibold, Benjamin uhttp://csl.arizona.edu/content/fuzzy-based-approach-dampen-emergent-traffic-waves06036nas a2200253 4500008004100000245005300041210005300094520525700147100002205404700001505426700002105441700002105462700003105483700002305514700002105537700001505558700002205573700002405595700002105619700001905640700002105659700002305680856007905703 2016 eng d00aDampening traffic waves with autonomous vehicles0 aDampening traffic waves with autonomous vehicles3 aIn congested traffic, minor disturbances or fluctuations in the velocity of a single vehicle may induce dynamically evolving traffic waves such as stop-and-go waves. These waves cause vehicles upstream to slow down or stop before accelerating back to the desired speed, resulting in increases in fuel consumption and risk of collisions. This work postulates that by intelligently controlling a small number (e.g., 1-5%) of autonomous vehicles (AVs) soon to be present in the traffic flow, it is possible to dampen or completely remove these speed fluctuations in the entire traffic stream. By only making small changes to the way the AV drives compared to human drivers near the dynamic wave, we can significantly improves the smoothness of the overall traffic flow, and reduces fuel consumption of all vehicles on the road. Due to the inherent instability of dense traffic flow, small disturbances in the speed of individual vehicles can generate large-scale disturbances in the traffic stream in the form of waves. Uncontrolled, these waves will propagate indefinitely until the traffic density decreases and the instability dissipates. This phenomenon was first experimentally demonstrated in the famous ring road experiment of Sugiyama et al., 2008. In that experiment, 22 vehicles were driven on a circular track to demonstrate that the uniform initial traffic flow (uniform speed and spacing) quickly devolves into a stop-and-go wave with vehicles at one side of the track at a complete standstill, while vehicles at the other side of the track are racing to keep up with the vehicle in front of them. To learn effective AV control strategies to dampen these traffic waves we must accurately simulate traffic in the specific conditions under which these waves arise. This will allow us to study how traffic responds to different control mechanisms implemented by the AV mathematically from a control prospective as well as in simulation. In this work, a microscopic car following model is used to simulate traffic with a mix of human-controlled and autonomous vehicles. We model human traffic flow using the combined optimal-velocity follow-the-leader (OV-FTL). This model is calibrated using trajectories of vehicles under human driving behavior such that the macroscopic quantities (average velocity, wave growth time, wave propagation speed) match those quantities observed in the Sugiyama experiment. Note that thus, the microscopic model is calibrated to reproduce real traffic waves, which is not commonly done in the traffic modelling community. Using linear stability theory, this model is shown to be unstable, since it has positive-valued eigenvalues. These manifest themselves in the form of stop-and-go waves when the calibrated model is used to simulated individual vehicle’s trajectories in time. In order to calibrate more realistic models of human drivers in dense, unstable, traffic conditions, field experiments are conducted using between 12 and 22 vehicles at the University of Illinois in Urbana, Illinois to re-create the traffic waves observed in the Sugiyama experiment, and probe the state space of traffic conditions under which such traffic instabilities will arise. This data is then used to calibrate more realistic models of human driving behavior that cover a broader range of traffic conditions. All vehicles used in the experiments are equipped with onboard diagnostics (OBD-II) scanners to record the vehicle’s velocity, engine speed, fuel rate, and fuel consumption throughout the experiment. This provides additional data that allows us to compare fuel consumption in traffic with stop-and-go waves to uniformly-flowing traffic. Furthermore, the trajectory of each vehicle is tracked using a 360-degree panoramic camera located at the center of the circular track. To begin, this research addresses the case of the 22-vehicle system recorded in the Sugiyama experiment and augment it by replacing one of the vehicles with an AV, which provides actuation in the system since it can be controlled to drive arbitrarily smoothly within the constraints set by the vehicle immediately in front of it. To apply linear stability theory, this augmented system is then linearized about an equilibrium traffic flow. We then use a feedback controller and pole placement to stabilize the system, and prevent traffic waves from emerging. Results from simulating the stabilized system in time indicate that a single AV using realistic control gains is able to dampen traffic waves in a 22-vehicle system without decreasing the average speed. The societal implications of this work are broad since most drivers experience delays and increased fuel consumption due to unstable and non-uniformly flowing traffic. While complete automation of the entire vehicle fleet may be many years away, in the short term, it is likely that some vehicles will be capable of driving autonomously in the near future. This research demonstrates then even with only a small percentage of vehicles driving autonomously, it is possible alter the traffic flow and prevent instabilities from arising. This results are lower fuel consumption and a shorter driving time not only for the autonomous vehicles, but all vehicles in the traffic stream.1 aStern, Raphael, E1 aWu, Fangyu1 aChurchill, Miles1 aWork, Daniel, B.1 aMonache, Maria, Laura Dell1 aPiccoli, Benedetto1 aPohlmann, Hannah1 aCui, Shumo1 aSeibold, Benjamin1 aHamilton, Nathaniel1 aHaulcy, R’mani1 aBhadani, Rahul1 aBunting, Matthew1 aSprinkle, Jonathan uhttp://csl.arizona.edu/content/dampening-traffic-waves-autonomous-vehicles01613nas a2200145 4500008004100000245005400041210005400095260001200149520114500161100002301306700001901329700001501348700002201363856008201385 2016 eng d00aRobust Control of Autonomous Vehicle Trajectories0 aRobust Control of Autonomous Vehicle Trajectories c07/20163 aIn this paper we describe a robust treatment of tracking trajectories with an autonomous vehicle. In employing autonomous behaviors for traffic control there will inevitably be disturbances introduced through model error, non-planar surfaces, sensor noise, and delay in both sensing and actuation. We describe how we address these issues through robust control techniques. The trajectories we follow include position and orientation as part of their specification: but the most interesting aspect of these trajectories is the time-varying description of the state. This is opposed to a traditional approach of following a trajectory at any speed (with expected error in all dimensions of the state vector), as long as the speed does not exceed a maximum value. However, for traffic control to reduce traffic waves, most of the dampening approaches are time-varying trajectories. With this in mind, it becomes necessary to consider the delay of following the reference trajectory, and how this may affect drivers in the flow. We include simulation data demonstrating the results, as well as data from a full-sized robotic Ford Escape.
1 aSprinkle, Jonathan1 aBhadani, Rahul1 aCui, Shumo1 aSeibold, Benjamin uhttp://csl.arizona.edu/content/robust-control-autonomous-vehicle-trajectories