ISSN: 2180-1053 Vol. 2 No. 1 January-June 2010 Adaptive Gain Controller Using Model Reference Adaptive Control Method Stability Approach for Road Vehicle Following System 55 AdAptive GAin Controller UsinG Model referenCe AdAptive Control Method stAbility ApproACh for roAd vehiCle followinG systeM M. r. sapiee1, A. noordin1 1Faculty of Electrical Engineering, Universiti Teknikal Malaysia Melaka, Locked Bag 1752, Pejabat Pos Durian Tunggal, 76109 Durian Tunggal, Melaka. Email: 1mohd.razali@utem.edu.my ABSTRACT In order to maintain stability and satisfy operating constraints, the control system on the following vehicle needs information about the motion of preceding vehicle. A one-vehicle look-ahead control strategy is proposed and will be investigated for this operation. A mathematical model for this control strategy is obtained and simulated. This paper describes the process of designing an adaptive gain controller for a road vehicle following system. This is done through simulations and is further discussed to find the effectiveness of the method. KEYWORDS: One-vehicle look-ahead control, model reference adaptive control, stability approach. 1.0 introdUCtion In general, Malaysian drivers tend to follow another vehicle closely. Hence, platoons or convoys appeared to develop rapidly. This normally happens when spacing between vehicles is close and there is no chance for the following vehicle to overtake the preceding vehicle. The vehicle at the back or the following vehicle will have to adjust its speed and spacing with respect to the preceding vehicle. If this is not taken into careful consideration, collision between them may occur. In order to avoid any collisions between the following vehicle and the preceding vehicle, it is necessary to maintain some safe distances between both vehicles at any speed. So, the following vehicle needs to have information regarding to the speed and the distance between them. In a normal driving, the driver of the following vehicle will estimate the information and adjust his vehicle’s speed and position to ISSN: 2180-1053 Vol. 2 No. 1 January-June 2010 Journal of Mechanical Engineering and Technology 56 have a nominal speed and a safe distance between his vehicle and the preceding vehicle. Nowadays, all the information does not need to be estimated. Sensors are available to measure the speed of the preceding vehicle and the position of the preceding vehicle with respect to the following vehicle. The information from the sensors will be used and processed by the following vehicle controller to produce the required speed and the safe spacing distance. This is where autonomous control can take place. The controller can automatically ensure safe distance based on the information obtained from the preceding vehicle. The autonomous controller on the following vehicle can activate the vehicle cruise control mode, where the driver does not need to hold the steering nor press the fuel pedal, and automatically apply the brake when necessary in order to ensure the safety of the vehicle. In developed nations, the autonomous concept leads to the Intelligent Vehicle Highway System (IVHS). As a vehicle enters the highway, his vehicle automatically takes-over the control of the vehicle while following the preceding vehicle. This feature also gives rise to steering less technology where during the autonomous control in action, the driver does not need to hold the steering wheel. All the driving tasks are taken care by the vehicle intelligent system. One of the autonomous features is the adaptive type control based on certain control strategy which gives rise to Adaptive Cruise Control (ACC). An ACC controlled vehicle will follow the front vehicle at a safe distance. A Model Reference Adaptive Control (MRAC) can be used in this type of control where the vehicle controller has the ability to adapt to the variation of speed and position of the preceding vehicle. 2.0 one-vehiCle looK-AheAd Control strAteGy When a vehicle follows another vehicle in front of it, a vehicle convoy or a vehicle following system is formed as shown in Figure 1. ISSN: 2180-1053 Vol. 2 No. 1 January-June 2010 Adaptive Gain Controller Using Model Reference Adaptive Control Method Stability Approach for Road Vehicle Following System 57 following vehicle front vehicle FIGURE 1 A vehicle following system The following vehicle can be controlled in such a way that it will maintain either the same speed to that of the immediate preceding vehicle, which is the front vehicle or maintain a safe distance in order to avoid collision between them. This is where the string stability plays an important role by having a string stable vehicle following system. The system is said to be stable if the range errors decrease as they propagate along the vehicle stream. In this control strategy, the controlled vehicle only refers the information from the preceding vehicle. So, the control system on the following vehicle needs information about the motion of preceding vehicles. Yanakiev, D. and Kanellakopoulos, I. used a simple spring-mass- damper system to demonstrate the idea of string stability and show the string-stability criterion for constant time-headway and variable time- headway policies. A mathematical modeling for this control strategy is shown in Figure 2 where i denotes the following vehicle and i-1 denotes the preceding vehicle. Vehicle i vehicle i-1 kp kv ix •• 1− •• ix ix 1−ix m i mi-1 FIGURE 2 Mathematical model of a vehicle following system control strategy Performing the mathematical modelling on only the following vehicle i and applying the Newton’s Second law results in the following Equation (1). Assuming unit mass for Equation (1) and taking Laplace transform, gives the following transfer function. ISSN: 2180-1053 Vol. 2 No. 1 January-June 2010 Journal of Mechanical Engineering and Technology 58 The transfer function in Equation (2) depends on the following spacing policy. The aim of this strategy is to maintain string stability for longitudinal motion within the vehicle following system or the vehicle convoy, particularly between a vehicle with a vehicle or between a vehicle with a following vehicle. This strategy is adopted in order to design a controller by investigating the following two policies. 2.1 spacing policy Researchers are trying to get close inter-vehicular spacing between vehicles, in order to have an effective following system. Spacing policy is defined as a rule that dictates how the speed of an automatically controlled vehicle must regulate as a function of the following distance. A control system should be designed such that it regulates the vehicle speed according to the designed spacing policy. There are three basic spacing policies employed by many researchers thus far. Those spacing policies are the fixed spacing policy, the constant time headway policy and the variable time policy. Only the first two will be discussed and investigated in this paper. 2.1.1 fixed spacing policy Under this policy, a fixed inter-vehicular spacing is implemented regardless of the vehicle’s speed. A well-known result states that it is impossible to achieve string stability in an autonomous operation when this spacing policy is adopted. This is mainly due to the relative spacing error that does not attenuate as it propagates down the string at all frequencies. Spacing error attenuation will only occur for frequencies above certain level. In addition, keeping the same fixed spacing at different convoy speed would risk the safety and comfort of passengers, especially when the vehicles are closely separated. Obviously, at higher speed, faster vehicle reaction time is needed in an emergency situation to avoid collision. Nevertheless, the fixed spacing policy can give guaranteed string stability if the front vehicle provides its information on its speed and or position to the rear vehicle in the convoy. This can be done through radio communication or the rear ISSN: 2180-1053 Vol. 2 No. 1 January-June 2010 Adaptive Gain Controller Using Model Reference Adaptive Control Method Stability Approach for Road Vehicle Following System 59 vehicle having sensors to detect the above two parameters. Equation (2) shown before is the transfer function for the fixed spacing policy. 2.1.2 fixed headway spacing policy This spacing policy keeps a fixed time interval, called time headway or headway, h, between the preceding vehicle and the following vehicle. It is a speed dependent policy where the inter-vehicular spacing will vary according to the preceding vehicle speed. At higher speed, vehicles will be separated in a greater distance but always maintains a fixed time interval between vehicles. Most researchers used this spacing policy in designing controllers to ensure string stability as this policy mimics the behavior of human drivers. As vehicle speed is increased, a human driver will keep a safe inter-vehicular spacing with the immediate preceding vehicle. The performance of the fixed headway spacing policy used in autonomous and cooperative vehicles following systems has been studied. It is found that there exists minimum possible fixed headway spacing before the string stability of a convoy collapses which is related to the actual dynamics of the vehicle. The effect of this fixed headway spacing policy is equivalent to the introduction of additional damping in the transfer function, which allows the poles of the transfer function to be moved independently from the zeros of the same transfer function. With the addition of the fixed headway spacing, Equation (2) then becomes with the control law developed as To simplify the control law and at the same time ensure stability, a pole-zero cancellation technique is chosen. This can be achieved by introducing the constraint ISSN: 2180-1053 Vol. 2 No. 1 January-June 2010 Journal of Mechanical Engineering and Technology 60 Thus, Figure 3 shows the pole-zero maps with pole and zero cancelling each other. Through the pole-zero cancellation technique, Equation (3) is thus reduced to which is a first order transfer function with a stable response with time delay. While kv must always be positive, there exists a pole which is always in the left hand side of the s-plane. Thus, the system is always stable for this spacing policy. Thus, h k 1 v (6) Figure 3 shows the pole-zero maps with pole and zero cancelling each other. Through the pole-zero cancellation technique, Equation (3) is thus reduced to v v 1i i 2 ks k sx sx sG (7) which is a first order transfer function with a stable response with time delay. While kv must always be positive, there exists a pole which is always in the left hand side of the s-plane. Thus, the system is always stable for this spacing policy. FIGURE 3 Pole-zero cancellation map 2.2 Inclusion of Vehicle Dynamics After proving that the fixed time headway policy is suitable to be adopted, a simplified vehicle dynamics model is introduced in order to mimic the actual vehicle internal dynamics. In this case, the external dynamics is not considered. In the simplified model, the internal dynamics is represented as a lag function i.e., the actual vehicle acceleration is obtained after a certain time delay . This is given by the relation in Equation (8). uaa (8) -2 -1.8 -1.6 -1.4 -1.2 -1 -0.8 -0.6 -0.4 -0.2 0 -1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1 Pole-Zero Map Real Axis Im ag in ar y A xi s pole and zero cancelling each other FIGURE 3 Pole-zero cancellation map 2.2 Inclusion of Vehicle Dynamics After proving that the fixed time headway policy is suitable to be adopted, a simplified vehicle dynamics model is introduced in order to mimic the actual vehicle internal dynamics. In this case, the external dynamics is not considered. In the simplified model, the internal dynamics is represented as a lag function i.e., the actual vehicle acceleration is obtained after a certain time delay . This is given by the relation in Equation (8). ISSN: 2180-1053 Vol. 2 No. 1 January-June 2010 Adaptive Gain Controller Using Model Reference Adaptive Control Method Stability Approach for Road Vehicle Following System 61 Equation (3) is modified to include the vehicle dynamics part and this gives a transfer function in Equation (9). Having designed the control strategy and by including the vehicle dynamics, the block diagram of the one-vehicle look-ahead vehicle following system is shown in Figure 4. Control strategy Vehicle dynamics vi-1 ai vi xi ui xi-1 FIGURE 4 Block diagram consisting of the control strategy and vehicle dynamics If τ is so small (as in an ideal vehicle) i.e. , then . Hence, the transfer function is reduced back to a second order transfer function. Equation (10) is then simulated by using MATLAB Simulink giving a speed input for h=1s and kp values of 0.5, 1, 2 and 3. Shown in Figure 5 and Figure 6 are the speed and acceleration responses, respectively. Further analysis of Figure 5 shows that kp value of 2 gives the best response. ISSN: 2180-1053 Vol. 2 No. 1 January-June 2010 Journal of Mechanical Engineering and Technology 62 FIGURE 5 Speed response of fixed headway with various kp values FIGURE 6 Acceleration response of fixed headway with various kp values 3.0 Model referenCe AdAptive Control An adaptive controller can modify its behavior in response to changes in the dynamics of a system and the character of any disturbance. It is a controller with adjustable parameter and a mechanism for adjusting the parameter. An adaptive control system consists of two loops, normal ISSN: 2180-1053 Vol. 2 No. 1 January-June 2010 Adaptive Gain Controller Using Model Reference Adaptive Control Method Stability Approach for Road Vehicle Following System 63 feedback loop with plant and controller and an adaptive parameter mechanism loop. Figure 7 illustrates the general structure of a Model Reference Adaptive Control (MRAC) system. FIGURE 7 General structure of an MRAC system The basic MRAC system consists of four main components: i) Plant to be controlled ii) Reference model to generate desired closed loop output response iii) Controller that is time-varying and whose coefficients are adjusted by adaptive mechanism iv) Adaptive mechanism that uses ‘error’ (the difference between the plants and the desired model output) to produce controller coefficient Regardless of the actual process parameters, adaptation in MRAC takes the form of adjustment of some or all of the controller coefficients so as to force the response of the resulting closed-loop control system to that of the reference model. Therefore, the actual parameter values of the controlled system do not really matter. 3.1 The Stability Approach The MRAC can be designed such that the globally asymptotic stability of the equilibrium point of the error difference equation is guaranteed. To do this, the Lyapunov Second Approach or stability approach is used. The term stability approach is used throughout this paper. It requires an appropriate Lyapunov function to be chosen, which could be difficult. This approach has stability consideration in mind and is also known as the Lyapunov Approach. In designing the MRAC controller, we would like the output of the closed-loop system (y) to follow the output of the reference model (ym). Therefore, we aim to minimise the error (e=y-ym) by designing a controller that has one or more adjustable parameters such that a certain cost function is minimized. e(t) ym(t) y(t) u(t) r(t) FIGURE 7 General structure of an MRAC system The basic MRAC system consists of four main components: i) Plant to be controlled ii) Reference model to generate desired closed loop output response iii) Controller that is time-varying and whose coefficients are adjusted by adaptive mechanism iv) Adaptive mechanism that uses ‘error’ (the difference between the plants and the desired model output) to produce controller coefficient Regardless of the actual process parameters, adaptation in MRAC takes the form of adjustment of some or all of the controller coefficients so as to force the response of the resulting closed-loop control system to that of the reference model. Therefore, the actual parameter values of the controlled system do not really matter. 3.1 the stability Approach The MRAC can be designed such that the globally asymptotic stability of the equilibrium point of the error difference equation is guaranteed. To do this, the Lyapunov Second Approach or stability approach is used. The term stability approach is used throughout this paper. It requires an appropriate Lyapunov function to be chosen, which could be difficult. This approach has stability consideration in mind and is ISSN: 2180-1053 Vol. 2 No. 1 January-June 2010 Journal of Mechanical Engineering and Technology 64 also known as the Lyapunov Approach. In designing the MRAC controller, we would like the output of the closed-loop system (y) to follow the output of the reference model (ym). Therefore, we aim to minimise the error (e=y-ym) by designing a controller that has one or more adjustable parameters such that a certain cost function is minimized. 4.0 AdAptive GAin Controller desiGn An adaptive gain controller is to be designed for the one-vehicle look- ahead control strategy with fixed time headway and vehicle dynamics by applying a Model Reference Adaptive Control (MRAC). This section presents a direct adaptive controller design which adapts the unknown vehicle parameter kp. The advantage of the adaptive approach is that unpredictable changes in the value of kp can be easily accommodated. Figure 8 on the other hand shows the one-vehicle look-ahead control strategy based controller with the gain parameter to be adjusted, kp is the gain adjustment mechanism to be designed in the control strategy. 4.0 ADAPTIVE GAIN CONTROLLER DESIGN An adaptive gain controller is to be designed for the one-vehicle look-ahead control strategy with fixed time headway and vehicle dynamics by applying a Model Reference Adaptive Control (MRAC). This section presents a direct adaptive controller design which adapts the unknown vehicle parameter kp. The advantage of the adaptive approach is that unpredictable changes in the value of kp can be easily accommodated. Figure 8 on the other hand shows the one-vehicle look- ahead control strategy based controller with the gain parameter to be adjusted, kp is the gain adjustment mechanism to be designed in the control strategy. FIGURE 8 A one-vehicle look-ahead controller with adjustable gain 4.1 The Stability Approach Design From the analysis of Figure 6, kp value of 2 gives the best response. So, it will be used in Equation (10) to give a reference model to be used in designing the adaptive gain controller. The vehicle dynamic has been included in the control law to form the plant. Plant: r ksk1s ks y pp 2 p (11) Reference model: r 2s3s 2s y 2m (12) 1 X''i 1 h Product 1 KvAdd2 Add1 Add 5 kp 4 Xi 3 X'i 2 X'i-1 1 Xi-1 Xi X'i X"i FIGURE 8 A one-vehicle look-ahead controller with adjustable gain 4.1 the stability Approach design From the analysis of Figure 6, kp value of 2 gives the best response. So, it will be used in Equation (10) to give a reference model to be used in designing the adaptive gain controller. The vehicle dynamic has been included in the control law to form the plant. Plant: ISSN: 2180-1053 Vol. 2 No. 1 January-June 2010 Adaptive Gain Controller Using Model Reference Adaptive Control Method Stability Approach for Road Vehicle Following System 65 Reference model: A closed-loop system with a controller has the following parameters: r(t) = Reference input signal u(t) = Control signal y(t) = Plant output ym(t) = Reference model output e(t) = Difference between plant and reference model output = y(t) - ym(t) In designing an MRAC using stability approach, the following steps should be followed: i) Derive a differential equation for error, e = y − ym that contains the adjustable parameter, . From Equations (11) and (12), after replacing, the differential equation becomes Substituting equations (13) and (14) into , thus Let , so that . Thus ii) Find a suitable Lyapunov function, usually in a quadratic form (to ensure positive definiteness). ISSN: 2180-1053 Vol. 2 No. 1 January-June 2010 Journal of Mechanical Engineering and Technology 66 The Lyapunov function, , is based on Equation (16). , where is positive definite. The derivative of V becomes where for stability must be negative i.e. . iii) Derive an adaptation mechanism based on such that e goes to zero. Therefore, The block diagram implementation is given in Figure 9 where is denoted by gamma in the simulation diagram and the red dotted line is the adaptive mechanism. iii) Derive an adaptation mechanism based on X,e,eV such that e goes to zero. reX 3 1 (18) Therefore, erk p (19) The block diagram implementation is given in Figure 9 where is denoted by gamma in the simulation diagram and the red dotted line is the adaptive mechanism. FIGURE 9 The stability approach adaptive gain controller simulation diagram. The MRAC stability approach adaptive gain controller design is then simulated again using MATLAB Simulink. Both the output of the system responses (y and ym) are shown in Figure 10 and Figure 11. Figure 10 shows a perfect model following output while Figure 11 on the other hand shows the acceleration response of y where it does not follow a sharp change in input acceleration. ui X'i Xi vehicle dynamics -K- gamma Xi-1 X'i-1 X'i Xi kp X''i control strategy s+2 s +3s+22 T ransfer Fcn Product 1 s Integrator2 1 s Integrator du/dt Derivative Add 1 X'i-1 y m e y yy FIGURE 9 The stability approach adaptive gain controller simulation diagram. The MRAC stability approach adaptive gain controller design is then simulated again using MATLAB Simulink. Both the output of the system responses (y and ym) are shown in Figure 10 and Figure 11. Figure 10 shows a perfect model following output while Figure 11 on the other hand shows the acceleration response of y where it does not ISSN: 2180-1053 Vol. 2 No. 1 January-June 2010 Adaptive Gain Controller Using Model Reference Adaptive Control Method Stability Approach for Road Vehicle Following System 67 follow a sharp change in input acceleration. FIGURE 10 Comparison of y and ym for speed from the stability approach FIGURE 11 Comparison of y and ym for acceleration from the stability approach It can be seen that the output response for plant y perfectly follows the reference model ym. It can be said that the system is a perfect model following system. The adaptive gain controller is again simulated but this time with gamma γ values of 0.1,0.01 0.001 and 0.0001 while the value of h is fixed at 1s. 0 20 40 60 80 100 120 140 160 180 200 0 5 10 15 20 25 Time (s) S pe ed ( m /s ) ym y 0 20 40 60 80 100 120 140 160 180 200 -1.5 -1 -0.5 0 0.5 1 1.5 Time (s) A cc el er at io n (m /s 2 ) ym y FIGURE 10 Comparison of y and ym for speed from the stability approach FIGURE 10 Comparison of y and ym for speed from the stability approach FIGURE 11 Comparison of y and ym for acceleration from the stability approach It can be seen that the output response for plant y perfectly follows the reference model ym. It can be said that the system is a perfect model following system. The adaptive gain controller is again simulated but this time with gamma γ values of 0.1,0.01 0.001 and 0.0001 while the value of h is fixed at 1s. 0 20 40 60 80 100 120 140 160 180 200 0 5 10 15 20 25 Time (s) S pe ed ( m /s ) ym y 0 20 40 60 80 100 120 140 160 180 200 -1.5 -1 -0.5 0 0.5 1 1.5 Time (s) A cc el er at io n (m /s 2 ) ym y FIGURE 11 Comparison of y and ym for acceleration from the stability approach It can be seen that the output response for plant y perfectly follows the reference model ym. It can be said that the system is a perfect model following system. The adaptive gain controller is again simulated but this time with gamma γ values of 0.1,0.01 0.001 and 0.0001 while the value of h is fixed at 1s. ISSN: 2180-1053 Vol. 2 No. 1 January-June 2010 Journal of Mechanical Engineering and Technology 68 FIGURE 12 Speed response with various gamma values for adaptive gain controller FIGURE 13 Acceleration response with various gamma values for adaptive gain controller Shown in Figure 12 and Figure 13 are the speed and acceleration responses for the adaptive gain controller, respectively. Further analysis of Figure 13 gamma value of between 0.01 and 0.001 gives the best response by almost fitting the reference model with smooth curve and with jerk of less than 5 m/s3. 0 20 40 60 80 100 120 140 0 5 10 15 20 25 Time (s) S pe ed ( m /s ) ym Gamma=0.1 Gamma=0.01 Gamma=0.001 Gamma=0.0001 0 20 40 60 80 100 120 140 -1.5 -1 -0.5 0 0.5 1 1.5 Time (s) A cc el er at io n (m /s 2 ) ym Gamma=0.1 Gamma=0.01 Gamma=0.001 Gamma=0.0001 FIGURE 12 Speed response with various gamma values for adaptive gain controller FIGURE 12 Speed response with various gamma values for adaptive gain controller FIGURE 13 Acceleration response with various gamma values for adaptive gain controller Shown in Figure 12 and Figure 13 are the speed and acceleration responses for the adaptive gain controller, respectively. Further analysis of Figure 13 gamma value of between 0.01 and 0.001 gives the best response by almost fitting the reference model with smooth curve and with jerk of less than 5 m/s3. 0 20 40 60 80 100 120 140 0 5 10 15 20 25 Time (s) S pe ed ( m /s ) ym Gamma=0.1 Gamma=0.01 Gamma=0.001 Gamma=0.0001 0 20 40 60 80 100 120 140 -1.5 -1 -0.5 0 0.5 1 1.5 Time (s) A cc el er at io n (m /s 2 ) ym Gamma=0.1 Gamma=0.01 Gamma=0.001 Gamma=0.0001 FIGURE 13 Acceleration response with various gamma values for adaptive gain controller Shown in Figure 12 and Figure 13 are the speed and acceleration responses for the adaptive gain controller, respectively. Further analysis of Figure 13 gamma value of between 0.01 and 0.001 gives the best response by almost fitting the reference model with smooth curve and with jerk of less than 5 m/s3. A one-vehicle look-ahead control strategy with fixed headway policy has been adopted in designing a controller to produce an output which can respond immediately to the change in input; in this case, the input ISSN: 2180-1053 Vol. 2 No. 1 January-June 2010 Adaptive Gain Controller Using Model Reference Adaptive Control Method Stability Approach for Road Vehicle Following System 69 is the speed with varying speed conditions. With normal controller, the response does not quite match perfectly with the input. With the introduction of the MRAC adaptive gain controller, the response can be made to follow the input by choosing a suitable reference model. Furthermore, using MRAC adaptive gain controller produces a smooth output as compared to the other one. 5.0 ConClUsion The adaptive gain controller tuning has been investigated using MRAC concepts through the stability approach. Simple adaptation law for the controller parameters has been presented assuming that the process under control can be approximated by a second order transfer function. The developed adaptation rule has been applied and simulated. The results obtained show the effectiveness of the technique. The resulting performance could be improved by a better choice of the length of the adaptation period. The stability approach is used to provide guaranteed nominal stability. However, the stability approach controller can only have very small gain with 1 is the maximum limit. A further limitation of the approach is the assumption of a structure for the nominal system. In this paper, a second order transfer function has been assumed resulting from the assumption of a very small time delay between the command signal and the vehicle dynamics as in ideal vehicle. 6.0 ACKnowledGMents The authors would like to acknowledge their gratitude to Faculty of Electrical Engineering, Universiti Teknikal Malaysia Melaka for providing the resources and support in the completion of this paper. 7.0 referenCes K. Pirabakaran, V. M. Becerra. (2001). Automatic Tuning of PID Controllers using Model Reference Adaptive Control techniques. IECON’01, The 27th Annual Conference of The IEEE Industrial Electronics Society. pp. 736-740. ISSN: 2180-1053 Vol. 2 No. 1 January-June 2010 Journal of Mechanical Engineering and Technology 70 Liang C. Y. and Peng H. (1999). Optimal Adaptive Cruise Control with Guaranteed String Stability’. Vehicle System Dynamics. Volume 31. pp 313-330. M. R. Sapiee, H. Selamat, A. Noordin, A. N. Jahari (2008). PI Controller Design Using Model Reference Adaptive Control Approaches For A Chemical Process. Proceedings of 2008 Student Conference on Research and Development (SCOReD 2008). 26-27 Nov. 2008. pp. 155-1 to 155-4. P.A. Cook and S. Sudin. (2003). Convoy dynamics with bidirectional flow of control information. 10th, IFAC Symposium on Control in Transportation Systems, Tokyo, Japan. 4-6 August 2003. pp. 433-438. P.A. Cook and S. Sudin. (2002). Dynamics of convoy control systems. 10th IEEE Mediterranean Conference on Control and Automation, Lisbon, Portugal. 9-12 July 2002. WP7-2. S. Sudin and P.A. Cook. (2003). Dynamics of convoy control systems with two- vehicle look-ahead strategy. International Conference on Robotics, Vision, Information and Signal Processing, Universiti Sains Malaysia, Penang, Malaysia. 22-24 January 2003. pp. 327-332. S. Sudin and P.A. Cook. (2004). Two-vehicle look-ahead convoy control systems. 59th, IEEE Vehicular Technology Conference, Milan, Italy. 17-19 May 2004. Volume 5. pp. 2935-2939. S. Sudin (2005). Dynamics and control of vehicle convoy systems. PhD Thesis. School of Electrical and Electronic Engineering, Faculty of Engineering and Physical Sciences, The University of Manchester. Yanakiev, D. and Kanellakopoulos, I. (1996). A Simplified Framework for String Stability Analysis in AHS. Proc. of the 13th IFAC World Congress. Volume Q. pp.177-182.