Transcription

A Review on Cooperative Adaptive Cruise Control (CACC)Systems: Architectures, Controls, and ApplicationsZiran Wang (presenter), Guoyuan Wu, and Matthew J. BarthUniversity of California, RiversideNov. 7, 2018 @ Maui, IEEE ITSC

sApplicationsDiscussions

Introduction

01 Introduction – From CC to ACC to CACC Cruise Control (CC):Vehicle maintains a steady speed as set by the driver Adaptive Cruise Control (ACC):Vehicle automatically adjusts speed to maintain a safe distance fromvehicle ahead Cooperative Adaptive Cruise Control (CACC)

01 Introduction – Cooperative Adaptive Cruise Control Take advantage of the Vehicle-to-Vehicle (V2V) and Vehicle-to-Infrastructure (V2I)communications Form platoons/strings and driven at harmonized speed with smaller time gap(D. Jia et al., 2016)

01 Introduction – Cooperative Adaptive Cruise Control Safer than human driving by taking a lot of danger out of the equation Roadway capacity is increased due to the reduction of inter-vehicle time gap Fuel consumption and pollutant emissions are reduced due to the mitigationof unnecessary stop and go, and aerodynamic drag of following vehicles(source: www.youtube.com/watch?v LljnfGXos4c)(Volvo)(TechAdvisor, 2013)

Architectures

02 Architectures – System StructurePerceptionPlanningActuationTwo sources: data fromHigh-level controller is developedLow level controller convertswireless safety unit andin MATLAB/Simulink and loadedthe target speed commandson-board sensorsin the vehicle using a dSpaceinto throttle and brake actionsMicroAutoBox

02 Architectures – System Structure

02 Architectures – Communication Flow TopologyDenote how information is transmitted among vehicles in a CACC vehicle string Predecessor-following Predecessor-leader following Two predecessor-following Two predecessor-leader following Bidirectional

Controls

03 Controls – Distributed ControlConsensus ControlOptimal ControlModel Predictive ControlH-Infinity ControlSliding Mode ControlDistributed consensus algorithms in the field of multi-agent system are applied toCACC systemsOptimal controllers for CACC are formulated as structured convex optimizationproblem with the objective to minimize energy consumption or travel timeA real-time optimization problem is solved to compute optimal acceleration anddeceleration commands to minimize energy consumptionH-infinity control can deal with modeling uncertainties and external disturbances,thus is widely studied to improve the robustness of CACC systemBesides uncertainties and external disturbances, sliding mode control is also widelyused to address string stability issue

03 Controls – Distributed Consensus ControlConverge to a desired locationArrive at their desired locations whilepreserving the desired formation shape

03 Controls – Distributed Consensus Control𝑥ሶ 𝑖 𝑡 𝑣𝑖 𝑡𝑣ሶ 𝑖 𝑡 𝑎𝑖𝑗 [𝑥𝑖 𝑡 𝑥𝑗 𝑡 𝜏𝑖𝑗 𝑡 𝑙𝑖𝑓 𝑙𝑗𝑟 𝑥𝑗ሶ 𝑡 𝜏𝑖𝑗 𝑡𝑔𝑡𝑖𝑗 𝜏𝑖𝑗 𝑡𝑏𝑖 ] 𝛾𝑎𝑖𝑗 𝑥ሶ 𝑖 𝑡 𝑥𝑗ሶ 𝑡 𝜏𝑖𝑗 𝑡𝑖 2, , 𝑛, 𝑗 𝑖 1𝑔𝑥𝑖 𝑡Longitudinal position of vehicle 𝒊 at time 𝒕𝑡𝑖𝑗Inter-vehicle time gap𝑥ሶ 𝑖 𝑡Longitudinal speed of vehicle 𝒊 at time 𝒕𝑙𝑖𝑓Length between GPS antenna to frontbumper𝑣ሶ 𝑖 𝑡Longitudinal acceleration of vehicle 𝒊 attime 𝒕𝑙𝑗𝑟Length between GPS antenna to rearbumper𝑏𝑖Braking factor of vehicle 𝒊𝛾Tuning parameter𝑎𝑖𝑗𝜏𝑖𝑗 𝑡𝑖, 𝑗 th entry of the adjacency matrixCommunication delay at time 𝒕

03 Controls – Distributed Consensus Control𝑥ሶ 𝑖 𝑡 𝑣𝑖 𝑡𝑣ሶ 𝑖 𝑡 𝑎𝑖𝑗 [𝑥𝑖 𝑡 𝑥𝑗 𝑡 𝜏𝑖𝑗 𝑡 𝑙𝑖𝑓 𝑙𝑗𝑟 𝑥𝑗ሶ 𝑡 𝜏𝑖𝑗 𝑡 𝛾𝑎𝑖𝑗 𝑥ሶ 𝑖 𝑡 𝑥𝑗ሶ 𝑡 𝜏𝑖𝑗 𝑡velocity consensusPredecessor following topology𝑔𝑡𝑖𝑗 𝜏𝑖𝑗 𝑡position consensus𝑖 2, , 𝑛, 𝑗 𝑖 1𝑏𝑖 ]

03 Controls – Distributed Consensus Control AssumptionEvery vehicle in the system is equipped with appropriate sensors Protocol 1: Normal platoon formation

03 Controls – Distributed Consensus Control Protocol 2: Merging and splitting maneuvers

03 Controls – Distributed Consensus Control Scenario 1: Normal platoon formation

03 Controls – Distributed Consensus Control

03 Controls – Distributed Consensus Control Scenario 2: Platoon restoration from disturbances

03 Controls – Distributed Consensus Control Scenario 3: Merging and splitting maneuvers

03 Controls – Distributed Consensus Control Scenario 3: Merging and splitting maneuvers

Applications

04 ApplicationsVehicle PlatooningCooperative Eco-DrivingCooperative MergingAutonomous IntersectionVehicles driven in a form of platoon/string with harmonized speed and constant timeheadwayVehicles collaborate with others to conduct eco-driving maneuvers along signalizedcorridorsVirtual CACC string can be developed to allow vehicles to merge in a cooperativemannerCollision-free intersection without traffic signal can be designed by CACC technology

04 Applications – Cooperative Eco-DrivingCEDLeaderCED VehicleCEDFollowerIn theV2IrangeEADOut ofthe V2IrangeIDMCACC

04 Applications – Cooperative Eco-Driving Only CED vehicles are classified into leaders and followers, whileconventional vehicles are not CED leaders conduct eco-driving maneuvers with respect to the trafficsignals through V2I communications CED followers follow the movements of CED leaders through V2Vcommunications

04 Applications – Cooperative Eco-Driving1. The vehicle’s longitudinal acceleration is controlled by the proposeddistributed consensus algorithm𝑎𝑟𝑒𝑓 𝛽 𝑑𝑔𝑎𝑝 𝑑𝑟𝑒𝑓 𝛾 (𝑣𝑝𝑟𝑒 𝑣𝑒𝑔𝑜 )𝑑𝑟𝑒𝑓 𝑚𝑖𝑛(𝑑𝑔𝑎𝑝 , 𝑑𝑠𝑎𝑓𝑒 )𝑑𝑔𝑎𝑝 𝑣𝑒𝑔𝑜 𝑡𝑔𝑎𝑝2. The estimated time-to-arrival should be updated all the time, in casethe CED follower cannot travel through the intersection during thegreen phase – in that case, the CED follower changes into CED leader

04 Applications – Cooperative Eco-DrivingThe simulation study is conducted based on the University Avenue corridor in Riverside, CA

04 Applications – Cooperative Eco-Driving

04 Applications – Cooperative Eco-DrivingSimulation setup and energy results

04 Applications – Cooperative MergingBenefits of cooperative on-ramp merging system Increase merging safety by applying V2X communications Increase traffic mobility by assigning vehicles into cooperative adaptive cruise controlstring before merging Reduce energy consumption by avoiding unnecessary speed changes

04 Applications – Cooperative MergingFollow thereference vehicleINFProcess data.Assign sequence.

04 Applications – Cooperative MergingMountain View, CA modeled in Unity3D environment

04 Applications – Cooperative MergingRamp modeled used to conduct simulation

04 Applications – Cooperative Merging

04 Applications – Cooperative MergingSimulation setting: 1 ramp vehicle, 6highway vehicles (already formed vehiclestring)

Discussions

05 DiscussionsThe realistic traffic network will introduce highly dynamic environment, includingchanging information flow topologies, varying workload distribution betweendifferent CAVs, and packet loss of V2V communicationsMethodologies need to be tested under all kinds of different conditionsand environments, and also for a rather long mileage. Since CACC systemsoften involves several CAVs, it would be difficult to conduct enough testsMaking new policies, updating roadside infrastructure, testing theproposed methods in real traffic cost a lot of money. To achieve apreferred penetration rate of CAVs in the application, the general publicalso need to spend money to purchase new vehiclesReliable ArchitectureReady-to-Market MethodologyReduce the Cost to Implement

Team MembersGuoyuan Wu, Ph.D.Adjunct Associate ProfessorElectrical EngineeringZiran WangMatthew J. Barth, Ph.D.Ph.D. CandidateProfessorMechanical EngineeringElectrical EngineeringWeb: www.me.ucr.edu/ zwang