Control Methods For Autonomous Vehicle Platoon System Invoking Swarm Intelligence

The increasing vehicle population in both Asia and other parts of the world leads to increasingly severe traffic congestion.The research on intelligent transportation systems has become a hot button issue since it can remit the traffic pressure effectively.As a branch of intelligent transportation system,autonomous vehicle platoon systems emerge at a historic moment.An autonomous vehicle platoon system can increase the capacity,enhance the safety,increase driving comfort,and reduce the fuel consumption and emission.Therefore,the research on autonomous vehicle platoon system contributes to the traffic congestion reduction,energy conservation and pollution reduction.So far,studies on autonomous vehicle platoon system focus on the stability control of longitudinal following problem,i.e.,the stability of the space between vehicles in the platoon system.The existing strategies cannot well solve the control issues when platoon system is of large scale,the number of vehicles is varying,and/or nonlinear factors and uncertain external disturbances are presented.An autonomous vehicle platoon system can be regarded as a mechanical swarm system.We therefore introduce the concept of swarm intelligence into control via the inspiration of biological swarm behaviors.In swarming species,complex global behaviors are achieved solely relying on simple strategies performed by each agent.Motivated by the exotic behaviors of swarming species,we conduct the study of an artificial mechanical swarm system consisting of multi-agents.We also explore the kinematic and dynamical characteristics of swarm system.Each agent in artificial mechanical swarm system consists of mechanical components.Hence,the motion of each agent is subject to the principles of mechanics.The artificial mechanical swarm system is designed as per biological swarm system and mimics the behaviors of biological swarm system.The kinematic model of swarm behavior could be described by a function with swarm properties.The swarm behavior could be achieved by treating kinematic model as the constraint which restrains the behavior of each agent.The closed-form constraint force which drives the system to meet the constraint is obtained via the Udwaida-Kalaba approach.Such constraint force is regarded as,from control design point of view,the ideal control to shape swarm behavior.Udwaida-Kalaba approach has very broad potential of applications.No matter whether the constraint is holonomic or nonholonomic,one can apply the Udwadia-Kalaba approach to establish the equation of motion for constrained system.This,however,does not completely solve the control design problem.Due to the presence of uncertainty,it is not practical in reality to utilize the obtained constraint force(which is based on the model)directly.In order to deal with the uncertainty,more research is needed,which is an emphasis of this thesis.By introducing swarm intelligence into the control of autonomous vehicle platoon system,we focus on the vehicle self-organization,distance and velocity stabilization problems in the presence of the uncertainty.Since the original state of system is one-side bounded when collision avoidance is taken into account,we convert the bounded state into a global one by a creative transformation.After embedding the function which describes the swarm behavior into platoon system,the dynamical model of autonomous vehicle platoon system is established.By creatively treating the kinematic model as a constraint,the closed-form constraint force(i.e.,ideal control)is obtained via Udwadia-Kalaba approach.Based on this,a class of(non-adaptive)robust controls for the following vehicles in platoon under known bounded uncertainty is proposed.The control guarantees not only collision avoidance(the vehicles never collide),but also compact formation(the total space occupied by the platoon system is compact),and stability(the actual space between every two adjacent vehicles is stable under the proposed control).Since the bound of the uncertainty is sometimes unknown and its improper selection may lead to extra control cost,we then propose a class of adaptive robust controls for autonomous vehicle platoon system under unknown uncertainty.The unknown uncertainty bound is assumed to be relevant to a function with known structure and unknown parameters.A leakage type of adaptive law is proposed to emulate the unknown parameters so that the estimation of uncertainty bound is obtained accordingly.Due to the leakage feature of the adaptive law,the adaptive parameters will not always increase.This helps to conserve the implementation cost.Based on this adaptive law,an adaptive robust control is proposed on the foundation of the constraint force.The resulting control exhibits the features of both robustness and adaptiveness.The major performances include collision avoidance,compact formation,and stability.Comparing to the robust control,the adaptive robust control possesses a smaller control effort.Aiming at an alternative description of uncertainty,an optimal fuzzy control approach is then proposed.By describing uncertainty using fuzzy set theory,the fuzzy autonomous vehicle platoon system is established.The uncertainty is assumed to lie within a fuzzy set.Such method is distinct from the probability theory and any other IF-THEN rule-based fuzzy methods.Combining thenovel fuzzy description of the uncertainty bound and adaptive law,a class of fuzzy adaptive robust controls is presented.A fuzzy-based performance index,consisting of transient control cost and average control cost,is proposed.The optimal choice of control design parameters is cast into an optimization problem.The global solution to this problem is unique and in closed-form.This optimization problem is believed to be solved in the most elegant way.The resulting optimal fuzzy adaptive robust control guarantees the autonomous vehicle platoon system two categories of performances.First,the deterministic performance(i.e.,uniform boundedness and uniform ultimate boundedness)is assured regardless of the uncertainty.Second,the fuzzy performance(the minimization of the performance index)is assured.