Design of adaptive cruise control for road vehicles using topographic and traffic information

Transcription

1 Preprints of the 19th World Congress The International Federation of Autoatic Control Design of adaptive cruise control for road vehicles using topographic and traffic inforation Péter Gáspár Balázs Néeth Institute for Coputer Science and Control, Hungarian Acadey of Sciences, MTA-BME Control Engineering Research Group, Hungary; Fax: ; Phone: ; E-ail: Institute for Coputer Science and Control, Hungarian Acadey of Sciences, Hungary; Kende u , H-1111 Budapest, Hungary; Phone: ; E-ail: Abstract: The paper proposes the design of the speed of road vehicles, which iniizes control energy and fuel consuption while keeping travelling tie and, oreover, considers the local traffic inforation to avoid conflicts in congestions. Topographic data and speed liits on the road are incorporated into the design of fuel efficient operation of the vehicle. Since the biased consideration of fuel consuption ay lead to the reduction of speed, the traffic flow in the surroundings of the vehicle ay be ipaired. Thus, the inforation about the local traffic is an iportant factor considering the wider transportation syste. In the paper the energy-efficient predicted cruise control strategy is presented, which is able to adapt to the otion of the surrounding vehicles. In this way a balance between the designed speed and the flow of the local traffic can be guaranteed. Keywords: adaptive cruise control, control, road preview, traffic inforation, optiization. 1. INTRODUCTION AND MOTIVATION The developent of an energy-efficient operation strategy has been in the focus of research and developent centers, suppliers and anufacturers. The purpose of the energyefficient operation is to design the speed of road vehicles, in which several factors are taken into consideration such as energy requireent, fuel consuption, road slopes, eissions and travelling tie. Consequently, the control ethods are based on ulti-objective optiization criteria. The advantage of these ethods is that they can be connected to the adaptive cruise control strategy. Several papers have been published in these topics. The optiization proble was handled by using a receding (sliding) horizon control approach in Hellströ et al. (1); Passenberg et al. (9). In Hellströ et al. (9) the predicted control approach was evaluated in real experients, based on the cobination of GPS signals and inforation about the road geoetry. Néeth and Gáspár (13a) proposed the design of speed for road vehicles based on road inclinations, speed liits, a preceding vehicle in the lane and traveling tie. Saerens et al. (13) proposed an eco-cruise control strategy, in which the ulti-criteria optiization between journey tie and fuel consuption was converted into a constrained fuel optiization task. The research has been conducted as part of the project TÁMOP- 4...A-11/1/KONV-1-1: Basic research for the developent of hybrid and electric vehicles. The Project is supported by the Hungarian Governent and co-financed by the European Social Fund. Several scenarios were presented by Rakha et al. (6) for the relationship between travel tie, energy and the eission of the vehicle. In Asadi and Vahidi (11) a predictive cruise control was proposed, which was able to consider upcoing traffic signal inforation to iprove fuel econoy and reduce traveling tie. The algorith used radar and traffic signals to optiize traveling speed. In the case of hybrid electric vehicles the road prediction is iportant to optiize battery recovery. The efficiency of terrain preview was deonstrated using siulation exaples in Zhang et al. (1). Abühl and Guzzella (9) presented the predictive reference signal generator ethod to axiize recuperated energy using the topographic profile of the future road segents and the corresponding average traveling speeds. In van Keulen et al. (1) a speed optiization ethod was detailed for heavy electric trucks. In the ethod the shape of the vehicle speed profile at a road segent was predefined and its paraeters were deterined during a nonlinear constrained optiization process. Schakel et al. (1) analysed the effects of cooperated adaptive cruise control on the ixed traffic flow. During the cooperation between vehicles the shockwaves in traffic can be daped efficiently. In this way the individually-controlled vehicles are able to guarantee a coordinated otion and iprove traffic flow. The control is also otivated by the design of a platoon control syste. A cooperative control strategy based on preview inforation, which initiates the change in speed for all vehicles in the platoon, was proposed by Copyright 14 IFAC 4184

2 Ala et al. (13). The ethod of control was extended to the design of the coon speed of the vehicles in the platoon in Néeth and Gáspár (13b). The goal was to deterine the coon speed at which the speeds of the ebers are as close as possible to their own optial speed. Although the efficiency of terrain preview consideration in cruise control has been proposed in several publications, a saller ephasis has been placed on traffic flow ipair caused by speed reduction. In ost papers, the traffic flow in the environent of the controlled vehicle is influenced during the journey. However, in the ethods listed above, the otion of the other vehicles on the road is not taken into consideration. For exaple, the driver of the vehicle is able to create a balance between energy/fuel saving and journey tie according to his own priorities. However, other drivers on the road have different priorities, which can lead to conflict, e.g. fast vehicles are held up by vehicles traveling in a fuel efficient fashion. The goal of the research is to design an optial control strategy which is able to adapt to the otion of the surrounding vehicles. In this way a balance between fuel econoy and the speed of vehicles on the road can be guaranteed. The cobination of the concept and the congestion proble leads to a coplex ulticriteria optiization task. The paper is organized as follows. The principles of the applied concept are briefly presented in Section. In Section 3 the interaction of strategy with the follower vehicle otion is forulated using optiization criteria. Section 4 presents the architecture of the design ethod incorporating the optiization processes. The efficiency of the ethod is illustrated through a siulation exaple in Section 5. Finally the contributions of the paper are suarized in Section 6.. PRINCIPLES OF LOOK-AHEAD CONTROL In the section the principles of the consideration of road conditions in speed profile design are briefly suarized. A detailed description of the ethod is found in Néeth and Gáspár (13a). The road ahead of the vehicle is divided unevenly, which is consistent with the topography of the road. In the ethod the vehicle is assued to be traveling in a segent fro the initial point to the first division point. The speed at the initial point is predefined and it is called original speed. The ai is to calculate the so-called odified speed at the sae initial point at which the reference speed of the first point can be reached by using a constant longitudinal force. This thought can be extended to the next segents and division points. In the case of n nuber of segents and n + 1 nuber of points as Figure 1 shows, n equations are forulated between the first and the end points. It is assued that the acceleration of the vehicle ay change in the different intervals, but within an interval the acceleration is constant. The otion of the vehicle is described using siple kineatic equations: s 1 = ξ ( ξ 1 ξ )/ ξ + ( ξ 1 ξ ) // ξ, where ξ is the speed of the vehicle at the initial point, n original reference velocities: vref F l 1 s 1 s vref1 od ifi ed reference velocity : ξ Fig. 1. Division of road α 1 s 3 α 1 α 4 vref vref3 vref4 vref5 vref6 vrefn ξ 1 is the speed of vehicle at the first point and s 1 is the distance between these points. F di disturbances considered in vehicle dynaics are divided in two groups: The first group is force resistance fro the road slope F di,r, which can be considered as a easured signal during the road slope easureent. The second group F di,o contains all of the other resistances such as rolling resistance and aerodynaic forces. The uneasured resistances F di,o can be approxiated by the square of ξ for the road sections as F di,o = A + T ξ, where A and T are constant vehicle paraeters. Based on the previous kineatic odeling fraework the speed of the vehicle at point i {1,,..., n} is written as: ξ i = ξ + i j=1 s j (F lj F dj ). Siilarly, the speed of the vehicle is forulated in the next n section points. Using this principle, a speed-chain which contains the required speeds along the way of the vehicle is constructed. At the calculation of the control force it is assued that additional longitudinal forces F li, i {,..., n} will not affect the next sections. It eans that always the actual F l1 control force ust be coputed and applied as oentary actuation. Thus, the predicted speed of the i th section point is the following: ξ i = ξ + i (s 1F l1 s j F dj ) (1) j=1 When the vehicle arrives at a speed liit, the vehicle speed ξ i ust reach this liit v ref,i. Therefore the following ai ust be achieved for all of the section points: ξ i vref,i. This defined condition is rearranged using the kineatical relationship and the for of disturbance. Consequently, the equations of the vehicle speeds at the section points are calculated in the following way: ξ vref,, () ξ + s 1F l1 s 1F d1,o vref,i + i s j F dj,r, (3) j=1 where i {1,,..., n}. Since the vehicle travels in traffic and it ay happen that it catches up with a preceding vehicle. Due to the risk of collision it is necessary to consider the speed of the preceding vehicle v lead : ξ v lead (4) In the next step prediction weights γ 1, γ,..., γ n are applied to (3). They represent the priority of the i th condition. Two additional prediction weights are applied: Q and W based on () and (4), respectively. The weights ust su up to one, i.e. γ 1 + γ γ n + Q + W = 1. (5) 4185

3 While the prediction weights γ i represent the rate of the road conditions, weight Q has an essential role: it deterines the tracking requireent of the current reference speed v ref,. By increasing Q the oentary speed becoes ore iportant while road conditions becoe less iportant. For exaple when Q = 1 the control task is siplified to a cruise control proble without any road conditions. By aking an appropriate selection of the weights the iportance of the road condition is taken into consideration. When equivalent weights are used the road conditions are considered with the sae iportance, i.e., Q = γ i, i {1,,..., n}. W represents the tracking of the speed v lead. When W = 1 only the tracking of the preceding vehicle is carried out. The deterination of W is based on the distance fro the preceding vehicle to iniize collision risk, see Néeth and Gáspár (13a). By using (3) and (4) and taking the weights into consideration the following forula is yielded: ξ + s 1(1 Q W )F l1 s 1(1 Q W )F d1,o = ϑ where value ϑ depends on the road slopes, the reference speeds and the weights ϑ = W vlead + Qvref, + γ i vref,i + n s i F di,r γ j. (7) In order to take the road conditions into consideration in the control design (6) is applied as a perforance of the controlled syste. Finally, a speed tracking proble is deduced, whose reference signal contains the predicted road inforation (road slopes, speed liits), such as: ξ λ (8) where paraeter λ is calculated in the following way based on the designed ϑ: λ = ϑ s 1 (1 Q W )( ξ + gsinα) (9).1 Optiization of the vehicle cruise control Equation (6) shows that the odified speed ξ depends on the prediction weights (W, Q and γ i ). By choosing these values the effects of road conditions can be tuned. The design of the vehicle speed profile poses two optiization probles, which are written in the following fors: Optiization 1: The longitudinal control force ust be iniized, i.e., F l1 in. Instead, in practice the Fl1 in optiization is used because of the sipler nuerical coputation. It leads to a quadratic optiization proble, which is written in the following for using (6): ˉF l1 = (β ( ˉQ) + β 1 ( ˉQ)ˉγ β n ( ˉQ)ˉγ n ) in (1) with the following constrains ˉQ, ˉγ i 1 and ˉQ + ˉγi = 1 W. This task is nonlinear because of the weights. At fixed Q weights the optiization task is solved by the transforation of the quadratic for into the linear prograing using the siplex algorith. Optiization : The difference between oentary speed and odified speed ust be iniized, i.e., v ref, ξ in. In this case the optial solution can be deterined j=i (6) easily, since the vehicle tracks the predefined speed if the road conditions are not considered. Consequently, the optial solution is achieved by selecting the weights in the following way: Q = 1 W and γi =, i [1, n]. The two optiization criteria lead to different optial solutions. In the first criterion the road inclinations and speed liits are taken into consideration by using appropriately chosen weights ˉQ, ˉγ i. At the sae tie the second criterion is optial if the inforation is ignored. In the latter case the weights are noted by Q, γ i. A balance between the perforances ust be achieved, which is based on a tuning of the weights. Several ethods can be applied in this task. In the proposed ethod two further perforance weights, i.e., R 1 and R, are introduced. Perforance weight R 1 ( R 1 1) is related to the iportance of the iniization of the longitudinal control force F l1 (Optiization 1) while perforance weight R ( R 1) is related to the iniization of v ref, ξ (Optiization ). There is a constraint according to the perforance weights R 1 + R = 1. Thus the perforance weights, which guarantee a balance between optiizations tasks, are calculated in the following expressions: Q = R 1 ˉQ + R Q = 1 W R1 (1 ˉQ W ) (11a) γ i = R 1ˉγ i + R γ i = R 1ˉγ i (11b) with i {1,.., n}. The equations show that prediction weights depend on R 1 linearly. Based on the calculated perforance weights the odified speed can be deterined by using (9). 3. CONSIDERATION OF THE MOTION OF THE FOLLOWER VEHICLE IN SPEED DESIGN Norally the driver sets weight R 1 based on his goals and requireents, thus he creates a balance between energy saving and travelling tie. However, a vehicle preferring energy saving ay be in conflict with other vehicles preferring cruising at the speed liit. Thus, an energyefficient vehicle ay decelerate the other vehicles on the road. Preferring weight R 1 leads to a non-optial otion for the traffic globally. In this section a weight calculation ethod which guarantees a balance between the energyefficient speed profile and the flow of the local traffic is proposed for R 1. Thus, the otion of the vehicle using the control and the otion of the follower vehicle are analyzed. 3.1 Predicting the speed of the vehicle using control The speed prediction of the vehicle using control is based on (3). In the following the uneasured resistances F di, is considered. Based on (7) the expression of ϑ can be rewritten as: ϑ =W vlead + (1 W )v ref, R 1(1 ˉQ W )vref, + + R 1 ˉγ i vref,i + R 1( s i F di,r ˉγ j ) = j=i =R 1 ˉϑ + v ref, (1 R 1 )(1 W ) (1) where ˉϑ contains the value of ϑ calculated with energyefficient prediction weights ˉQ, ˉγ i. 4186

4 Fro (8) the reference speed λ is calculated based on the predicted road inforation. It shows that through Q and ϑ weight R 1 plays an iportant role in the calculation of the reference speed. Moreover, the predicted values of the weights γ i also depend on R 1, see (11). Fro (6) and (7) the square of the reference speed ξ is calculated in the following for: λ =R 1 ˉϑ + v ref, (1 R 1 )(1 W )+ s 1 R 1 ( 1 ˉQ W ) ( ξ + gsinα) =R 1 ( ˉϑ s1 (1 ˉQ W ) + v ref, (1 R 1)(1 W ) =R 1ˉλ + v ref, (1 R 1)(1 W ) (13) Fro (3) and (13) the predicted estiated speed of the vehicle at section point n is ξ n = ξ + s 1F l1 s 1F d1,o s i F di,r = λ (1 s 1T ) + s 1F l1 s 1A s i F di,r = R 1 N 1 + N (14) According to (14) the predicted speed at point n is independent of v ref,n. For exaple when R 1 = the predicted speed at point n ust be v ref,n. However, using (14), the value ξ i depends only on the oentary speed liit v ref,, while future speeds v ref,i do not influence it. Based on (14) the defined reference speed at section point n ust be odified in the following way: ξ n = (R 1 N 1 + N ) R 1 + (1 R 1 )vref,n (15) Consequently, the estiated speed at section point n is ξ n = (R 1 N 1 + N ) R 1 + (1 R 1 )vref,n (16) which depends on the weight R 1. In the forula N 1 is independent of the section points ahead, while N contains the road grade inforation of each section. 3. Predicting the otion of the follower vehicle Hereinafter it is necessary to deterine the criterion of the safety distance between the vehicle using the lookahead control and the follower vehicle. It requires the estiation of the otion prediction of the follower vehicle. The controlled vehicle oves fro point ξ to ξ 1, whose distance is s 1 while the traveling tie is Δt 1. Meanwhile the follower vehicle oves fro point η to η 1. In the optiization ethod it is assued that although the acceleration of the vehicle ay change in the different intervals, within an interval acceleration is constant. Thus, the traveling tie in the first interval is expressed as Δt 1 = s 1 /( ξ 1 +λ), where λ and ξ 1 are fro equations (13) and (15), respectively. The traveling tie between points ξ 1 and ξ is expressed siilarly as Δt = s /( ξ + ξ 1 ). In the estiation of the follower vehicle several assuptions are considered. First, the preceding vehicle has inforation about the speed and acceleration of the follower vehicle ( η, η ) and the oentary distance between the vehicles e. Second, the follower vehicle accelerates evenly until it reaches the speed liit. When the follower vehicle reaches the speed liit v ref,j it does not accelerate further, thus in the oncoing sections the predicted speeds of the vehicle are v ref,j,..., v ref,n. Note that the optial speed of the preceding vehicle is continuously re-designed, thus the otion of the follower vehicle ust be predicted. In the following, based on the inforation η, η, e the otion of this vehicle ust be calculated in every sections in which the traveling tie is Δt i, i = {1... n}. Until the follower vehicle reaches the speed liit, i.e., k < j, the distance of the vehicle is the following: ( k ) k η k = η Δt i + η (Δt i ) (17) where k [1,..., j 1]. When the follower vehicle reaches the speed liit at section j the equation is the following: ( j 1 ) η l = η j 1 l Δt i + η (Δt i ) + (v ref,i Δt i ) (18) where l [j,..., n]. After this section the speed of the follower vehicle is considered v ref,l. 3.3 Safety distance criterion Now the safety distance between the vehicle using the lookahead control and the follower vehicle ust be guaranteed. The safety distance s safety is assued to be predefined. The controlled vehicle intends to use the energy-efficient predicted cruise control, while the follower vehicle ais to keep the speed liit. Thus, the control strategy is odified in such a way that the otion of the follower vehicle is taken into consideration. A possible ethod is to odify weight R 1 during the journey and create a balance between the designed speed and the required speed of the follower vehicle. The ai of this section is to develop a ethod for the re-design of weight R 1. The criterion of the safety distance is based on the otion of the vehicles. During the journey in every section the distance between the two vehicles ust be guaranteed by the following inequalities: ξ i + e η i s safety, i {1,,.., n} (19) where ξ i is the predicted displaceent of the controlled vehicle, e is the oentary distance between the vehicles and η i is the predicted displaceent of the follower vehicle. It is necessary to find the axiu of weight R 1, which guarantees the inequality constraints (19). Note that an increase in R 1 induces longer journey tie. Therefore R 1 can be liited by the driver using a predefined bound R 1,ax. The optiization criterion of the safety cruising is forulated as follows: ax R 1 () [;R 1,ax] such that the following conditions are satisfied: j s i + e η j s safety, j {1,..., n} (1) In these inequalities η j depends on weight R 1. The result of the optiization R 1,opt is used in the calculation of the prediction weights Q and γ i. Based on the prediction weights and equation (9) the reference speed λ of the controlled vehicle is coputed. The optiization procedure () is perfored in each step, thus weight R 1 is rewritten continuously according to the current local traffic inforation. i=j 4187

5 th IFAC World Congress 4. ARCHITECTURE OF THE SPEED PROFILE DESIGN METHOD In practice the solution of the optiization processes (1), () ay require a great deal of coputation effort. However, the constrained quadratic optiization proble (1) is reforulated to a linear prograing task. In this way the coputation of ˉQ, ˉγ i requires less tie. In the case of the other optiization () the solution of the previous coputation step R 1,old is applied as initial value. The new solution R 1,new is searched in the interval [ax(r 1,old α, ), in(r 1,old + α, R 1,ax )] with n = 1 points and α =.1. Note that R 1,ax is set by the driver. Its default value is R 1,ax = 1. Both optiization (1) and () are solved with saple tie.5s. The purpose of this procedure is to guarantee that the coplexity of the optiization ethod is reduced and, thus, the ethod can be applied in practice. Figure illustrates that the vehicle receives inforation about the otion of preceding vehicle (distance and speed) and the follower vehicle (distance, speed, acceleration). Moreover, the road inclination α i and speed liit v ref,i data are also known. Since the route of the vehicle is considered as known, the loading of this inforation requires the easureent of the current position, e.g. using GPS. speed liit data vehicle data easureent vehicle data road data roadside unit easureent local traffic inforation Fig.. Counication and inforation flow vehicle data Ebnre and Herann (1) provided a survey of the future counication possibilities in autootive and traffic control. Nuevo et al. (1) presented a coputer vision-based approach to tracking surrounding vehicles and estiating their trajectories. An extension of adaptive cruise control with traffic inforation considering vehicle-to-roadside and vehicle-to-vehicle counication was proposed in Kesting et al. (7). Festag et al. (8) cobined vehicle-to-vehicle counication and vehicleto-roadside sensor counication to prevent accidents and assist investigations. 5. SIMULATION RESULTS In this section the proposed speed design ethod is deonstrated through a siulation exaple using highcoplexity vehicle dynaic software CarSi. In the scenario a aneuver is considered, in which a controlled vehicle with the presented ethod overtakes slower preceding vehicles on the highway (controlled). The overtaking aneuver is carried out by using an energy-efficient ethod. At the sae tie another vehicle equipped by an conventional adaptive cruise control drives onto the highway and accelerates to reach the speed liit and also begins an overtaking aneuver (follower). Thus, there is a conflict between the vehicles caused by the decreased distance between the controlled and follower vehicles during their aneuver. In the following exaple the efficiency of conflict handling based on the proposed control strategy is presented. The terrain characteristics of the road are illustrated in Figure 3(a). This road contains downhill sections, whose inclinations are 5% and 3%. The energy-efficient cruising of the vehicle requires the reduction of vehicle speed before the downhill sections. The speed liit on the highway is 13k/h, which is reduced to 11k/h before the second inclination. The speed profiles of the controlled vehicle with a control strategy and the follower vehicle are shown in Figure 3(b). The follower vehicle drives onto the busy highway at 8k/h and accelerates dynaically to reach the speed liit 13k/h. The controlled vehicle is oving at approxiately 11k/h, since the acceleration effect of the predicted slope is considered. In the exaple R 1,ax =.75 is set by the driver. Note that this is the tuning paraeter. Consequently, the perforance weight changes in the interval R 1 R 1,ax ethod. At the beginning of the siulation perforance weight R 1 is chosen.75, therefore the control force F l1 is iniized, see Figures 3(e) and 3(f). However, the distance between the vehicles is reduced, which ay cause a conflict. In order to hold the safety distance s safety the priority of R 1 ust be reduced, see Figure 3(f). At 45 sec the overtaking aneuver of the controlled vehicle is finished, thus R 1 is reset to.75. The reduction of R 1 results in the increase of vehicle speed in the first siulation case during the increase of R. Therefore F l1 is increased, see Figure 3(f). It leads to higher fuel consuption, since it is necessary to avoid a hazardous conflict. In the other siulation the otion of follower is not considered by the controlled vehicle. Figure 3(d) shows the speed profile of the vehicles. Since the controlled vehicle considers ainly the predicted terrain characteristics R 1 =.75, the interdistance is decreased continuously, see Figure 3(c). Although in this case F l1 is iniized during the entire road, the otion of the controlled vehicle caused a traffic conflict. The follower vehicle ust significantly reduce its speed in its lane, which is unacceptable, see Figure 3(d). Besides, the interdistance is decreased below s safety, which is also dangerous on the highway. The energy consuption of the controlled vehicle using the presented strategy is shown in Figure 3(g). It is copared with a vehicle without traffic inforation and another vehicle without strategy. As long as R 1 =.75, the energy consuption of the vehicle with traffic inforation and the vehicle without that is the sae, see Figure 3(e). When R 1 is reduced control energy consuption increases significantly, because terrain characteristics are less considered. Thus, the tendency of energy consuption 4188

6 Road characteristics () Distance () R 1 Energy (kj) (a) Terrain characteristics (c) Interdistance of vehicles (e) Perforance weight R 1 5 without (g) Control energy consuption Speed (k/h) follower (b) Speed profile - with traffic inforation Speed (k/h) follower (d) Speed profile - without traffic inforation Force (N) without (f) Control force F l1 without Station () (h) Traveling tie Fig. 3. Siulation results of the overtaking aneuver is close to the vehicle without strategy. In the presented siulation scenario the energy consuption of the presented ethod is approxiately 75% of an uncontrolled vehicle. Figure 3(h) illustrates the traveling tie values of the vehicles. Although energy saving of the lookahead strategy is considerable, the traveling tie of the vehicle increases. Without strategy the journey tie is 87 sec, while the vehicle with strategy considering traffic inforation travels 3 in 9 sec. The tie increase of the proposed strategy is 3%. The results show that the energy consuption is significantly higher than the tie lost. 6. CONCLUSIONS The paper has proposed a speed design ethod which can be built in the adaptive cruise control strategy. In the ethod several factors such as energy requireent, road slopes, traveling tie, road disturbances and speed liits have been taken into consideration and a ulti-objective optiization procedure has been fored. Since the vehicle is part of the transportation syste, this energy-efficient cruise control strategy has been coordinated with the otion of the surrounding vehicles. In the design ethod perforance weight R 1 has been tuned in order to keep the safety distance between vehicles. The siulation exaple has shown that by considering the predicted speed of the other vehicles conflict events can be avoided. REFERENCES Ala, A., Martensson, J., and Johansson, K. (13). Look-ahead cruise control for heavy duty vehicle platooning. Proc. 16th IEEE Annual Conference on Intelligent Transportation Systes, The Hague. Abühl, D. and Guzzella, L. (9). Predictive reference signal generator for hybrid electric vehicles. Vehicular Technology, IEEE Transactions on, 58(9), Asadi, B. and Vahidi, A. (11). Predictive cruise control: Utilizing upcoing traffic signal inforation for iproving fuel econoy and reducing trip tie. Control Systes Technology, IEEE Transactions on, 19(3), Ebnre, A. and Herann, R. (1). A self-organized radio network for autootive applications. 8th World Congress on Intelligent Transportation Systes, Sydney, Australia. Festag, A., Hessler, A., Baldessari, R., Le, L., Zhang, W., and Westhoff, D. (8). Vehicle-to-vehicle and road-side sensor counication for enhanced road safety. 15th World Congress on Intelligent Transport Systes. Hellströ, E., Åslund, J., and Nielsen, L. (1). Horizon length and fuel equivalents for fuel-optial control. Advances in Autootive Control, Munich, 1 6. Hellströ, E., Ivarsson, M., Åslund, J., and Nielsen, L. (9). Lookahead control for heavy trucks to iniize trip tie and fuel consuption. Control Engineering Practice, 17(), Kesting, A., Treiber, M., Schönhof, M., and Helbing, D. (7). Extending adaptive cruise control to adaptive driving strategies. Transportation Research Record,, Nuevo, J., Parra, I., Sjoberg, J., and Bergasa, L. (1). Estiating surrounding vehicles pose using coputer vision. In 13th IEEE Conf. on Intelligent Transportation Systes, Néeth, B. and Gáspár, P. (13a). Design of vehicle cruise control using road inclinations. International Journal of Vehicle Autonoous Systes, 11(4), Néeth, B. and Gáspár, P. (13b). Optiised speed profile design of a vehicle platoon considering road inclinations. IET Intelligent Transport Systes. Passenberg, B., Kock, P., and Stursberg, O. (9). Cobined tie and fuel optial driving of trucks based on a hybrid odel. European Control Conference, Budapest, 1 6. Rakha, H., El-Shawarby, I., Arafeh, M., and Dion, F. (6). Estiating path travel-tie reliability. IEEE Intelligent Transportation Systes Conference, Toronto, Canada, Saerens, B., Rakha, H., Diehl, M., and den Bulck, E.V. (13). A ethodology for assessing eco-cruise control for passenger vehicles. Transportation Research Part D, 19, 7. Schakel, W., Van Are, B., and Netten, B. (1). Effects of cooperative adaptive cruise control on traffic flow stability. In 13th IEEE Conf. on Intelligent Transportation Systes, van Keulen, T., de Jager, B., Foster, D., and Steinbuch, M. (1). Velocity trajectory optiization in hybrid electric trucks. In Aerican Control Conference, Zhang, C., Vahidi, A., Pisu, P., Li, X., and Tennant, K. (1). Role of terrain preview in energy anageent of hybrid electric vehicles. IEEE Transactions on Vehicular Technology, 59(3),