Test-track Testing - Combined Longitudinal and Lateral Control Design

5. Combined Longitudinal and Lateral Control Design

6.1 Test-track Testing

The system implemented on Taiwan iTS-1 was firstly tested at the Automotive Research and Testing Center (ARTC) with respect to different weather and lighting conditions. The missions of automated lane-keeping with varying velocities on 2 testing-tracks are summarized in Table 6-1. Some events like rain which can not occur usually are denoted as N/A.

TABLE 6-1. Testing Conditions of Different Environmental Sets.

Sunny Cloudy Night Rainy

Weather Testing Track

CDTT 0 ~ 145 km/h 0 ~ 145 km/h 0 ~ 90 km/h N/A

NVHSTT 0 ~ 130 km/h 0 ~ 120 km/h 0 ~ 90 km/h N/A CDTT: Coast down test track.

NVHSTT: Noise vibration & harshness surface test track.

Lane-keeping mode was initially evaluated on the test track at ARTC, and the experiments of regulation were undertaken to examine the stability of the overall system at various speeds.

Initially the system was tested without FGS compensation. In this case, the automated steering control oscillated when the velocity was increased to around 70 km/h. When the vehicle continued accelerating, the steering action was untamed, worsening the system. Figure 6-1(a) shows sampled experimental results under these testing conditions. In the case of FGS compensation, the vehicle could be automatically steered in the lane of CDTT while the velocity was increased to around 145 km/h, as shown in Fig. 6-1(b). The vehicle also steered well on the NVHSTT with a harsh and vibration surface at a velocity of around 130 km/h, as shown in Fig. 6-1(c). Besides irregular surface, a segment of the NVHSTT only has single-side lane-marking. The performance of lane-keeping was indicated by the lateral offset

at a look-ahead distance (Ld =15 m) measured by the vision system. Thus, one can well imagine that the vehicle tracks the centerline of roads correctly in spite of varying vehicle velocities and vibration surface of the road. The control action of SW (in the right bottom plot) is small even the vehicle was accelerated up to a high velocity. This kind of action is conformed to a human-driver behavior.

Although not shown, the night test for the automated lane-keeping system also has been carried out with the maximum velocity 90 km/h, without any street lamps except the front vehicle lamp.

Fig. 6-1(a). The experimental results without FGS on the CDTT (straight lane with flat surface).

Fig. 6-1(b). The experimental results with FGS on the CDTT (straight lane with flat surface).

Fig. 6-1(c). The experimental results with FGS on the NVHSTT (straight lane with irregular surface including single-side lane marking segment).

6.2. Expressway/Highway Testing

Fig. 6-2. The snapshot of the experiment on expressway.

The experiments on freeway/highway have two stages. In the prior stage, only the throttle is ready to be used in the longitudinal system. The experiments are done at Expressway No.

68 (Chutung-Nanliao segment, Hsinchu, Taiwan) under real-traffic environment. Two vehicles were used to test the system, as shown in Fig. 6-2. The preceding-vehicle drives manually in simulating real traffic conditions (unpredicted cut-in/exist-out and speed-up/speed-down). The subject-vehicle (Taiwan iTS-1) is defaulted at hybrid LK and CC mode. Once a valid-target is detected, the switch to hybrid LK and ACC mode will be automatically done. The subject-vehicle adaptively adjusts its speed to maintain a safety-distance which is determined by the following-velocity. If forward traffic is clear or the subject-vehicle reaches its target speed, CC mode will be activated again.

The sample history of the vehicle following experiments without the employment of the adaptive detection area is depicted in Fig. 6-3. Here the headway time σ is set as 1 second, which is the minimum requirement of ISO 15622. Figure 6-3(d) shows the throttle input which is normalized from 0.13 (pedal fully released) to 0.82 (pedal fully pressed). Fig. 6-3(c) displays the desired velocity (dashed line) and current velocity (solid line) of the controlled vehicle. The desired headway distance (broken line) and the measured distance (solid line) from the range finder are shown in Fig. 6-3(b). The desired acceleration (with the bounded af max of 2 m/s²) calculated from the supervisory control is depicted in Fig. 6-3(a).

As evident in Fig. 6-3(b), there are some apparent pulse signals (illustrated in the circled area) which exist at the signal of the inter-vehicle distance measured from the range finder.

This is because at this instant, the preceding vehicle is undetected by the range finder due to the fact that the forward vehicle moving outside of detected area (target-missing). This

situation occurs when the two vehicles move on curves with an approximate radius of 400 m, and especially when the following vehicle is driven near the outside edge while the preceding vehicle is driven near the inside edge of the road. In Fig. 6-3(a), it can be seen that the calculated acceleration fluctuates dramatically. In this situation, the system switches to the higher velocity cruise control mode which causes the command of the throttle voltage to become erratic, as shown in Fig. 6-3(d). Consequently, for a vehicle velocity of 80-60-80 km/h, the initial detection angle of ±2° is deficient when compared to the required detection angle in the adaptive detection maneuver (±4.6°−5.9°, refer to Table 4-1 in Chapter 4).

Although the human driver can apply the brake to avoid the sudden acceleration of the controlled vehicle, such an action may not be desired due to the simultaneous actuation of both the throttle and the brake.

Fig. 6-3. The sampled history of vehicle following experiments without adaptive HDA maneuver (Dashed line: reference signal; solid line: real signal).

Fig. 6-4. The sampled history of experiments involving the transition between CC and ACC mode (Dashed line: reference signal; solid line: real signal).

Figure 6-4 depicts the sample history of experiments which involved the employment of the adaptive detection area. In contrast to Fig. 6-3(b), there was no pulse signal in Fig. 6-4(b).

This indicates that no target-missing detection occurred while the vehicles moved over the same curves with a radius of approximately 400 m. Initially the controlled vehicle moved with a velocity of 65 km/h, and one vehicle ahead is detected at 25 m, but it is outside of the desired operation range (18 m for the headway time σ = 1 s). Once the controlled vehicle accelerates to nearly 80 km/h and the preceding vehicle is within the operation range of 22 m at time = 10 s as shown in Fig. 6-4(b), the system adjusts the throttle to execute the desired velocity command for the vehicle following control. During the steady-state vehicle following (from 10 s to 50 s), not only does the subject vehicle track the desired velocity successfully as shown in Fig. 6-4(c), it also maintains the desired headway distance (in Fig. 6-4(b)) with respect to the preceding vehicle. It should be noted that the vehicle velocity is not fixed but it varies within the range of [70, 80] km/h. The maximum error in the safety-distance tracking is

0.6 m. When the preceding vehicle changes to the neighboring lane at 60 s, the automation system identifies the condition as one in which there is no vehicle ahead. Thus, the following vehicle automatically switches to the velocity cruise mode and accelerates to the pre-selected velocity of 88 km/h. As illustrated in Fig. 6-4(c), after 60 s point the velocity of the controlled vehicle converges smoothly to the reference velocity profile.

Fig. 6-5. Performance of ACC mode (fixed-distance-tracking) in a real traffic environment.

The experiment for two-car platoon-control scenario, namely fixed-distance-tracking of ACC mode, was demonstrated in the real traffic environment. The experimental results are shown in Figure 6-5. According to traffic condition (traffic-jam), a preceding-vehicle drives at varying velocity (around 70 km/h in the middle plot). The subject-vehicle (initial velocity at 62 km/h) accelerates to about 74 km/h automatically, and then decelerates to 70 km/h for maintaining a fixed headway-distance 10 m (at 12 s in the top plot). In real situation, it is arduous for a human driver in the preceding vehicle to manually drive at a fixed velocity.

Therefore, the subject-vehicle will adjust its throttle degree (in the bottom plot) slightly to maintain a fixed inter-vehicle distance 10 m. A good tracking is shown in spacing (in the top

plot of Fig. 6-5). Without braking assistance, the maximum intra-spacing deviation is 1 m only.

In the second stage, the brake actuation is added to assist the system in the vehicle longitudinal motion control. As a result, the subject-vehicle can manage throttle and brake to keep the reference distance or velocity against irregular road conditions such as non-flat, up-gradient, downhill level, and slippy or rough surface. Firstly, we demonstrate the reference velocity tracking, namely CC mode, at velocities from 20 to 90 km/h. The experimental results of CC mode are quantified at different speeds in Table 6-2 which shows the performance of the regulation control. ev represents the error between the real speed and the desired speed, | ev |avg is the absolute value of the mean error, and | ev |max is the absolute value of the mean error.

TABLE 6-2. Performance of CC mode at different velocities.

Speed | ev |avg (km/h) | ev |avg (%) | ev |max (km/h) | ev |max (%)

20 km/h 0.2037 1.02 0.6018 3.01

30 km/h 0.3313 1.11 0.6717 2.24

40 km/h 0.1503 0.38 0.6174 1.54

50 km/h 0.1086 0.22 0.3991 0.8

60 km/h 0.1450 0.24 0.4532 0.76

70 km/h 0.1913 0.27 0.6936 0.99

80 km/h 0.1815 0.23 0.5571 0.70

90 km/h 0.1751 0.19 0.6167 0.69

The operation from ACC mode to CC mode is the same to the prior stage due to only throttle actuation is required. Therefore, we demonstrate the scenario that the subject-vehicle operates from CC mode to ACC mode. Experimental results of CC to ACC mode are shown in Fig. 6-6. Initially the subject-vehicle speed is at 90 km/h for CC mode. In the same way, the headway time is chosen as 1 s. The headway distance is the solid signal which represents the relative distance between two vehicles, and the safe distance is shown by the dash-dot signal in the top figure. A preceding-vehicle appears in the front of the subject-vehicle at 14 s.

Since the headway distance suddenly drops, the system activates the brake control such that the headway distance can converge to the safe distance. At 18 s and 32 s, the

preceding-vehicle goes with higher speed so that the headway distance grows than the safe distance. In these two periods, the subject-vehicle intends to change to CC mode. However, at 27 s and 53 s, the relative distances become shorter than the safe distance. Thus the subject-vehicle decreases its velocity to keep the safe distance again. It can be seen that no matter the preceding vehicle’s speed changes severely and frequently, the system can automate the throttle and brake pedal to drive the subject-vehicle to maintain the safe distance and the desired speed.

Fig. 6-6. Experimental results of CC mode switching to ACC mode.

The time response of the process in ACC mode is shown in Fig. 6-7. The headway time is set as 1 s again. In the period of [0, 8] s, the system actuates the brake pedal such that the subject-vehicle decelerate its velocity to maintain the safe distance with respect to the preceding-vehicle. After 8 s, the preceding-vehicle runs at varying speeds within [60, 80]

km/h and the subject-vehicle keeps well the safe distance. It can be seen that actuating on the throttle during [10, 60] s, if necessary, is enough to assure fine tuning of the headway distance fitting the safe distance. After 60 s, the deceleration from 80 to 60 km/h requires applying brake control. This is similar to human driving condition: the driver applies the brake at the

end of the maneuver only, when it is strictly necessary. Otherwise, the driver just releases the throttle pedal since the vehicle can be decelerated by means of engine brake and tire-road resistance.

Fig. 6-7. ACC mode operation with throttle and brake actuation.

The performance for the LK mode in real traffic environments is shown in Fig. 6-8, where the subject-vehicle’s velocity is determined by CC or ACC mode. We find that more lateral offset (in the right-top graph) exists in the real traffic environments since road-curvature is an unknown external input. But, the measured lateral offset at a look-ahead distance is still within 0.8 m. Besides, the error between the measured and the command SW signals (in the bottom plot) is small enough to keep the subject-vehicle driving along the centerline.

Moreover, low lateral acceleration (within ±0.2g in the middle plot) indicates driving comfort is ensured.

Fig. 6-8. Performance of LK mode operation under real-traffic environment in Highway No.

The persuasive performance for the LK mode in real traffic environments is shown in Fig.

6-9. The subject-vehicle is moving at velocity 70 km/h across crucial curve, where the permissible highest velocity is 50-60 km/h. The curvature data is obtained from the vision system, and the maximum value is about 0.003 (≈ 300 m radius). As can be seen, the curve is with continuous changing and the maximum curvature approaches 1/200 m^-1. We find that

more lateral offset (in the top plot) exists in this situation. It can be concluded that the proposed FGS in the lateral control system reveals well capability of the curvature changing.

The steering control from the DSP board is almost equal to that of dSPACE MABX, and the SW motor follows the controlling command well to keep the subject-vehicle driving along the centerline of road. Moreover, low lateral acceleration (within ±0.2g in the middle plot) indicates that driving comfort is ensured.

Fig. 6-9. LK mode operation under crucial curves in Expressway No. 68.

Figure 6-10 shows the frame grabbed from the vision system in the operation of LC mode.

The lane-marking data will be lost due to the fact that the vision system cannot catch lane-marking in the lane-to-lane transition, as shown in the circled two plots of Fig. 6-10.

When the vehicle moved to an adjacent lane, the vision system re-caught the lane-marking and thus the system returned to LK mode. The exampled lane-change control at speed around 56 km/h is shown in Fig. 6-11. The LC mode started at 12 s, and the SW was then controlled by the lane-change controller instead of the lane-keeping controller. During the period of lane-change (from 12 s to 18 s), initially the lateral offset will increase since the vehicle steered to the boundary of road. While the vision system cannot catch the lane-marking, the lateral offset data hold a maximum value of 200 cm (between 14 s and 16 s). After the lane-change command was completed at 18 s, the SW returned to the lane-keeping controller and the subject-vehicle kept the centerline of road. Here the command signal from dSPACE MABX is in LK mode and it provides the comparison with the controlling of LC mode.

Figure 6-11 depicts the turn-left LC mode. The turn-right LC mode is shown in Fig. 6-12, and the velocity of subject-vehicle approximates 70 km/h. The overall lane-change maneuver is complete within 5~6 s which is according to our lane-change controller as designed in Chapter 3.

Fig. 6-10. Lane-marking detection in the lane-change scenario from the display of vision system.

Fig. 6-11. The transition between LK mode and turn-left LC mode.

Fig. 6-12. Turn-right LC mode operation with the velocity 70 km/h.

To improve the free lane-change’s feasibility, early-switch allows the lane-keeping control to intervene before LC mode completes. As shown in Figs. 6-13, once the vision system re-catches the neighboring lane-markings, the lane-keeping controller takes over the steering control to steer the subject-vehicle to the centerline of road. The intervention of LK mode is

illustrated in the red circle area, and it can be seen that the period of open-loop lane-change is shortened (about trisection) such that the disturbances from road surface and level can be reduced. In addition, the lateral acceleration during the lane-change is within the 0.2 g even the early-switch is applied. The required passenger comfort can be achieved through LK mode to LC mode, and back to LK mode.

Fig. 6-13(a). Turn-right LC mode at 62 km/h with early-switch maneuver.

Fig. 6-13(b). Turn-left LC mode at 67 km/h with early-switch maneuver.

6.3. Urban-road Testing

(a) Low speed CC mode. (b) Stop-and-go mode.

Fig. 6-14. The in-vehicle view for stop-and-go maneuver in the Nanliao Harbor Park.

In Fig. 6-14, the in-vehicle view represents typical traffic conditions at low speed in urban-road environment. Figure 6-14(a) represents that no preceding-vehicle in the front of the subject-vehicle and the system activates CC mode initially. The subject-vehicle will keep the reference velocity which can be chosen by the driver or the maximum velocity of urban-road. If a preceding-vehicle with lower speed appears in the same lane and finally stops ahead, the subject-vehicle reduces its speed to approach the preceding-vehicle gradually until it comes to a halt at a safe distance behind the preceding one. When the preceding-vehicle restarts moving, the subject-vehicle also moves and keeps a safe headway distance at any speed. Figure 6-14(b) represents the typical traffic jam scenario in which the vehicle continuously stopping and starting. The system adapts the subject-vehicle’s speed to follow the preceding vehicle which is driven manually, stopping when necessary and keeping a safe distance even when circulating at low speeds.

As shown in Fig. 6-15, the initial condition of this scenario is that both the preceding- and subject-vehicles are stationary in the same lane with over 50 m apart and facing in the same direction. Here the maximum speed of CC mode is set as 30 km/h, 2 s for the headway time, and 4 m for the minimum stopping distance which is the typical length of a vehicle. The subject-vehicle starts driving along its lane at 5 s, with initial CC mode. As the subject-vehicle approaches to a stopping preceding-vehicle, the headway distance decreases, and the system switches to the stop-and-go mode, adjusting the speed to keep a safe distance.

As the headway distance decreases, the throttle degree is also gradually released. Around 28 s,

the headway distance begins to be smaller than the safe distance, and thus the desired speed is less than the vehicle speed. At this point, the system applies the brake to slow down the vehicle, and the vehicle continues reducing speed until its headway distance reaches the minimum stopping distance. The second graph shows that the subject-vehicle stops at 4 m behind the preceding-vehicle from 32 s to 38 s. The preceding-vehicle then starts moving, followed by the subject-vehicle. This throttle/brake control behavior is very similar to human driving: The driver accelerates until a front car appears in the driving lane, then releases the degree on the throttle to slightly reduce speed and, if this reduction is not enough, applies the brake until the car stops without collision to the front car. This experiment reproduces the case of typical stop-and-go situation: the preceding vehicle is stationary and the following vehicle approaches the preceding one at high speed; this situation is very common at the tail end of a traffic jam and causes a lot of rear-end crashes.

0 10 20 30 40 50 60

Fig. 6-15. Low speed CC mode switches to stop-and-go mode.

0 10 20 30 40 50 60 70 80 90

Fig. 6-16. Stop-and-go mode of the supreme 30 km/h.

This scenario is the realization of a stop-and-go automatic car-following. Figure 6-16 indicates the reaction of the controlled vehicle when the preceding one carries out successive stopping and restarting. The controlled vehicle tracks the preceding one very accurately and respects the headway targets, 2 s for headway time, and 4 m for minimum headway distance by using both pedals when necessary. As shown in the top figure, the controlled vehicle stops at 4 m behind the preceding vehicle at 27 s, 54 s and 75 s. While the preceding vehicle restarts, the controlled vehicle always tracks the safe distance to move. This experiment reflects the situation that the controlled vehicle tracks preceding one which stops and starts alternatively, as in a traffic jam. As speeds and distances are low, the rear-end crash risk decreases, but

在文檔中自動車輛駕駛縱向暨橫向控制設計 (頁 92-0)