Adaptive Virtual Tow Bar and Transition of Control : A truck driving simulator study
report
Background : Truck platooning, with trucks being virtually connected, is getting more and more attention. Truck platooning offers the potential for substantial fuel savings while allowing the truck driver in the platoon to take a rest. However, at the current state of technology, truck drivers are still required to be alert and ready to intervene if required. How safe this situation of platooning is depends partly on whether the driver is able to take over control when requested. Under normal conditions, a transition back to the driver is requested by the system and the truck driver is provided with sufficient time and the situation is not too time critical.
Under other conditions, a driver may need to respond to other traffic soon after taking over control, for instance by means of braking. Quite some studies have been performed about transitions of control from automated to manual driving, however studies with professional truck drivers in various platooning situations are scarce.
Objective : The aim of this study was to investigate how long it takes before truck drivers take back manual control after a system warning if they can choose their own moment of getting back control.
Methods : We studied various scenarios, in which drivers either had to monitor the surroundings, work with a tablet or keep their eyes closed. Besides the response times, we also studied the quality of the driving behaviour right after taking back control in order to investigate whether drivers were actually ready to take back control in normal conditions and also in conditions in which they needed to brake as a response to a braking lead vehicle. 22 professional truck drivers took part in a truck driving simulator experiment. An automated motorway truck platooning system was simulated that allowed the participant to hook on to a lead truck and follow automatically at close distance with hands off the steering wheel and feet off the pedals. All participants made 9 drives, including a manual truck driving condition to get a baseline condition. After hooking on to the platoon, they drove in automated platooning mode until the system requested them to take back control. After getting a warning, they could press a button on the steering wheel to indicate that were ready after which they got back control. Response times were recorded, as well as the quality of driving behaviour after the transition and responses to a lead vehicle braking action. Also, situational awareness and acceptance were studied, as well as eye movements, feet, hand and body position as well as input from wearables measuring heart beat and arm movements.
Results : In case of voluntary take-overs, with truck drivers choosing their own moment to take back control after being requested by the system, response times vary quite substantially per condition and per driver. When drivers were asked to monitor the surroundings while platooning, mean response times were around 2.5 seconds, with not so much variance between drivers. For the condition in which drivers were working with a tablet, these response times doubled to around 5.5 seconds, with increasing variance. For the condition in which drivers had their eyes closed, mean response times were a little over 6 seconds, with a large variance and the slowest response times being over 16 seconds. When drivers had been platooning with shorter headways, they took somewhat more time to take back control.
After having taken back control, the driving performance was measured and compared to manual truck driving behaviour. Under normal driving conditions, we found that overall performance was comparable to their normal driving behaviour. However for the condition in which drivers had been working with a tablet, there seems to be some negative after-effect of platooning on lateral performance, although effects are relatively small and seem to disappear over time. In the conditions in which drivers have to respond to a braking lead truck after taking back control, we see adequate responses to the lead truck although the minimum time to collision is sometimes quite low. The wearables did not deliver reliable results since there was a very low correlation between the two wrist bands that were used. Automatic analyses of eye movement data, hand, feet and body position showed to be quite complex due to the fact that drivers sometimes obstructed the video images with body or arms, and that sometimes short glances were done to the road that were not automatically detected by the Smart Eye camera. In general, participants were rather positive about the system, with a score of 7 out of 10. The majority of drivers would like to have this system in their truck, even though the trust of the system in the simulator was higher than their trust if they imagined using this on the real road.
Conclusion : Drivers that were instructed to monitor the surroundings while truck platooning have short take-over times and lower variability in response times than drivers using a tablet or having their eyes closed during platooning. Remember that in this study, drivers could indicate themselves whether they were ready to take back control, so they could take more or less time. Drivers take more time to get back control when they have been platooning at shorter headways. Apparently truck drivers are aware that they had been out of the loop for a while and are aware of the involved risk in short following distances. There were large individual differences in response times between drivers, with large differences within one condition between drivers, but also with large differences between the different conditions. The large variability in response times is probably also due to the individual differences in body, hand and feet position, and in the fact that people were holding a tablet, were sometimes wearing reading glasses or changed the seating position of their chair. This is behaviour we could see on the video images, and more detailed analyses of the video images will be done in order to find further explanatory variables for differences in response times.
Under other conditions, a driver may need to respond to other traffic soon after taking over control, for instance by means of braking. Quite some studies have been performed about transitions of control from automated to manual driving, however studies with professional truck drivers in various platooning situations are scarce.
Objective : The aim of this study was to investigate how long it takes before truck drivers take back manual control after a system warning if they can choose their own moment of getting back control.
Methods : We studied various scenarios, in which drivers either had to monitor the surroundings, work with a tablet or keep their eyes closed. Besides the response times, we also studied the quality of the driving behaviour right after taking back control in order to investigate whether drivers were actually ready to take back control in normal conditions and also in conditions in which they needed to brake as a response to a braking lead vehicle. 22 professional truck drivers took part in a truck driving simulator experiment. An automated motorway truck platooning system was simulated that allowed the participant to hook on to a lead truck and follow automatically at close distance with hands off the steering wheel and feet off the pedals. All participants made 9 drives, including a manual truck driving condition to get a baseline condition. After hooking on to the platoon, they drove in automated platooning mode until the system requested them to take back control. After getting a warning, they could press a button on the steering wheel to indicate that were ready after which they got back control. Response times were recorded, as well as the quality of driving behaviour after the transition and responses to a lead vehicle braking action. Also, situational awareness and acceptance were studied, as well as eye movements, feet, hand and body position as well as input from wearables measuring heart beat and arm movements.
Results : In case of voluntary take-overs, with truck drivers choosing their own moment to take back control after being requested by the system, response times vary quite substantially per condition and per driver. When drivers were asked to monitor the surroundings while platooning, mean response times were around 2.5 seconds, with not so much variance between drivers. For the condition in which drivers were working with a tablet, these response times doubled to around 5.5 seconds, with increasing variance. For the condition in which drivers had their eyes closed, mean response times were a little over 6 seconds, with a large variance and the slowest response times being over 16 seconds. When drivers had been platooning with shorter headways, they took somewhat more time to take back control.
After having taken back control, the driving performance was measured and compared to manual truck driving behaviour. Under normal driving conditions, we found that overall performance was comparable to their normal driving behaviour. However for the condition in which drivers had been working with a tablet, there seems to be some negative after-effect of platooning on lateral performance, although effects are relatively small and seem to disappear over time. In the conditions in which drivers have to respond to a braking lead truck after taking back control, we see adequate responses to the lead truck although the minimum time to collision is sometimes quite low. The wearables did not deliver reliable results since there was a very low correlation between the two wrist bands that were used. Automatic analyses of eye movement data, hand, feet and body position showed to be quite complex due to the fact that drivers sometimes obstructed the video images with body or arms, and that sometimes short glances were done to the road that were not automatically detected by the Smart Eye camera. In general, participants were rather positive about the system, with a score of 7 out of 10. The majority of drivers would like to have this system in their truck, even though the trust of the system in the simulator was higher than their trust if they imagined using this on the real road.
Conclusion : Drivers that were instructed to monitor the surroundings while truck platooning have short take-over times and lower variability in response times than drivers using a tablet or having their eyes closed during platooning. Remember that in this study, drivers could indicate themselves whether they were ready to take back control, so they could take more or less time. Drivers take more time to get back control when they have been platooning at shorter headways. Apparently truck drivers are aware that they had been out of the loop for a while and are aware of the involved risk in short following distances. There were large individual differences in response times between drivers, with large differences within one condition between drivers, but also with large differences between the different conditions. The large variability in response times is probably also due to the individual differences in body, hand and feet position, and in the fact that people were holding a tablet, were sometimes wearing reading glasses or changed the seating position of their chair. This is behaviour we could see on the video images, and more detailed analyses of the video images will be done in order to find further explanatory variables for differences in response times.
Topics
TNO Identifier
575816
Publisher
TNO
Collation
55 p. (incl. appendices)
Place of publication
Soesterberg