Cons nsen ensus sus-based based co cooperat perative ive co cont ntrol rol ap approach roach ap applied lied to to ur urban an tr traffic affic ne netwo twork rk Antonio Artuñedo, Raúl M. del Toro, Rodolfo Haber {antonio.artunedo, raul.deltoro, rodolfo.haber}@car.upm-csic.es Centre for Automation and Robotics (UPM-CSIC) www.car.upm-csic.es 3rd International Electronic Conference on Sensors and Applications 15 – 30 November 2016 ECSA 2016 November 2016 1
Outline • Introduction • Proposed solution in a simulated environment • Modeling • Consensus-based cooperative control • Simulation (Open & closed loop) • Results & conclusions 3rd International Electronic Conference on Sensors and Applications November 2016 2
Introduction Current smart cities research aims to the integration of urban subsystems for the anticipation and control of daily situations and unexpected events in order to succeed under complex and potentially unstable conditions. Overall performance of the city is determined by the dynamic behavior of coupled physical subsystems which have different domains or timing aspects. One of the main challenges is the necessary cooperation among different entities such as vehicles or infrastructure systems and exploit the information available through networks of sensors deployed as infrastructures for smart cities. 3rd International Electronic Conference on Sensors and Applications November 2016 3
Introduction The increasing number of sensors, actuators, communication systems and low cost computation already deployed in cities, enable new applications that can go beyond specific systems and cover different urban systems and scenarios. In this work an algorithm for cooperative control of urban subsystems is applied in order to provide solutions for mobility related problems in cities. 3rd International Electronic Conference on Sensors and Applications November 2016 4
Proposed solution in a simulated environment Goal: Network of units for adaptive traffic lights TLC 1 TLC 2 control cycles • Improve performance of TLC 3 TLC 4 urban traffic networks, in ε specific regions of the city, Adaptive TLC TLC cycle based on air pollution Intersections network ξ adaptation traffic state information. Cyber world Δ u x Communication network Scenario based on: Traffic • Emission control scheme Mobility domain. suggested by Andó et. al. [1] Information service Road traffic Traffic subsystem providing pollution data control subsystem Pollution Pollution monitoring Air quality monitoring Other pollution stations of the city sources Pollution subsystem Physical environmet [1] B . Ando, S. Baglio, S. Graziani, E. Pecora, and N. Pitrone, ʺA predictive model for urban air pollution evaluationʺ, in Instrumentation and Measurement Technology Conference, 1997. IMTC/97. Proceedings. Sensing, Processing, Networking., IEEE , 1997, pp. 1056‐1059 vol.2. . 3rd International Electronic Conference on Sensors and Applications November 2016 5
Modeling: DEVS (Discrete Event Systems Specification) • It enables specification of basic components Pollution and how they are monitoring ξ connected together: Eo • atomic models, input coupled DEVS: TLC_network ports, changing states, Other pollution output ports, couplings. u 1 sources x 1 TLC1 ε 1 • Atomic models: Ev • Traffic-light control unit u 2 x 2 (TLC), TLC2 ε 2 • Pollution-monitoring Traffic system x 3 • Traffic system (i.e. road u 3 TLC3 ε 3 network, vehicles, traffic lights, etc.), x 4 • Other pollution sources TLC4 ε 4 u 4 • Coupled models: TLC network 3rd International Electronic Conference on Sensors and Applications November 2016 6
Modeling: DEVS (Discrete Event Systems Specification) • It enables specification classdef am_TLC < atomic %% Description of basic components Pollution % Adapt. Traffic Control Subsys. model and how they are monitoring ξ %% Superclass connected together: % |atomic| Eo • atomic models, input %% Class Methods coupled DEVS: TLC_network ports, changing states, %% Inherited Properties Other pollution %% User Defined Properties output ports, couplings. u 1 sources %% Ports x 1 TLC1 ε 1 • Atomic models: %% States in s Ev %% • Traffic-light control unit u 2 properties (Access = public) x 2 (TLC), TLC2 ε 2 accTflow = [0 0]; • Pollution-monitoring Traffic accMflow = []; system ... x 3 • Traffic system (i.e. road u 3 TLC3 end ε 3 network, vehicles, traffic methods lights, etc.), function obj = am_TLC(name,inistates,elapsed) x 4 • Other pollution sources ... TLC4 ε 4 u 4 end • Coupled models: TLC ... network end end 3rd International Electronic Conference on Sensors and Applications November 2016 7
Consensus-based decision-making Consensus: fundamental problem in the study of cooperative control for distributed multi-agent coordination. This approach deals with a set of systems each pursuing its own objectives as well as their common goals, employing communications between them. Why consensus? sensus? It’s proposed in the literature as an SoS cooperative-control paradigm to extract greater benefits from the constituent systems of an SoS [2]. Applications: cooperative control of vehicles, robots and rovers, wireless- sensor networks, traffic-optimization and control problems in urban environments [2] T. Nanayakkara, F. Sahin, and M. Jamshidi, Intelligent control systems with an introduction to system of systems engineering : CRC Press, 2010. 3rd International Electronic Conference on Sensors and Applications November 2016 8
Consensus-based cooperative control 1. Graph definition: TLC1 TLC2 TLC4 TLC3 2. Representing system dynamics by a consensus state variable – estimation of pollutant concentration at each intersection 𝜁 𝑗 𝑙 + 1 = 𝜁 𝑗 𝑙 + 𝛽 𝑗 𝜊 𝑙 − 𝑜 + 𝛾𝑦 𝑗 𝑙 − 𝑛 + 𝛿 Δ 𝑣 𝑗 Overall city Measured total number Control action in %: Current value pollution & of vehicles & relational change of traffic light of consensus intersection factor to intersection cycle & relational factor variable contribution factor emission to local emissions 3rd International Electronic Conference on Sensors and Applications November 2016 9
Consensus-based cooperative control 3. Consensus-based control law design Δ 𝑣 𝑗 𝑙 = − 1 𝛽 𝑗 𝜊 𝑙 − 𝑜 + 𝛾𝑦 𝑗 𝑙 − 𝑛 + 𝜇 𝑏 𝑗𝑘 𝜁 𝑗 − 𝜁 𝑘 𝛿 𝑘∈𝑂 𝑗 Feed-forward action Feed-forward Consensus action related to local action related to based on consensus pollution data local traffic data state of neighbors Note: control action is restricted to a variation of ±50% over the initial value, to avoid large dissimilarities with pre-defined traffic-light cycle lengths. 3rd International Electronic Conference on Sensors and Applications November 2016 10
Open loop scenario simulation Based d on an urban-li like ke road ad network work: • 4 signalized traffic intersections & fixed traffic-light cycles • Vehicles circulate following random routes. Tools ls: • SUMO microscopic traffic simulator • MatlabDEVS toolbox NO x emissions (AVG. for 20 secs of the Traffic queues at intersections (AVG. vehicle whole scenario) queues for 20 secs at every approach) 3rd International Electronic Conference on Sensors and Applications November 2016 11
Closed loop scenario simulation • Same scenario than open loop simulation • Parameters of control system are specified in section 2.3 of the paper (pp. 4-5) Vehicle queues at intersections (AVG. vehicle queues for 20 secs at every approach) 3rd International Electronic Conference on Sensors and Applications November 2016 12
Simulation results Diff fference ces s KPI (>100 00 scen enar ario io Open en-loo loop Closed-loo oop rel elati tive ve to simul ulat ations ions) open en-loo oop Vehicle queues μ 13,4815 12,0382 10,70 % 1 𝑢 𝑔 1. ‖𝑦‖ 𝑒𝑢 𝑢 𝑔 −𝑢 𝑡 max 15,0661 13,6345 9,50 % 𝑢 𝑡 Global pollution μ 2,3879·10 -4 2.3791·10 -4 0,37 % 1 𝑢 𝑔 ‖𝜊‖ 𝑒𝑢 2. 𝑢 𝑔 −𝑢 𝑡 min 2,2732·10 -4 2,1910·10 -4 3,62 % 𝑢 𝑡 The effect of balancing consensus variables in every TLC produces a global reduction of vehicle queues 3rd International Electronic Conference on Sensors and Applications November 2016 13
Conclusions • Discr crete ete ev even ent syste tem m spec ecific ificatio ation n (DEV EVS) S) modeling paradigm permitted operations with systems of a different nature and temporal behavior. • Consen ensus sus-based based control ol algorithms can be applied to the specific problems of traffic optimization. • KPIs KPIs and simulations showed that the number of vehicles in queue decreased, while consensus state variable at each intersection tended towards a common value, demonstrating the validity of the proposed solution. 3rd International Electronic Conference on Sensors and Applications November 2016 14
Thank you Antonio Artuñedo, Raúl M. del Toro, Rodolfo Haber {antonio.artunedo, raul.deltoro, rodolfo.haber}@car.upm-csic.es Centre for Automation and Robotics (UPM-CSIC) www.car.upm-csic.es 3rd International Electronic Conference on Sensors and Applications November 2016 15
Recommend
More recommend
