Cooperative Adaptive Cruise Control Human Factors Study: Experiment 2—Merging Behavior

CHAPTER 4. DISCUSSION

The goal of this study was to explore drivers’ abilities to merge with CACC platoons. Each participant completed four experimental merges in one of the following three different manners:

• Merging with non-CACC vehicle into a left dedicated lane without CACC platooning and varying vehicle gaps.
  
  
• Merging with CACC vehicle into the middle of a CACC platoon without speed assistance.
  
  
• Merging CACC vehicle into a CACC platoon with longitudinal speed assistance.

In addition to merging behavior and performance, workload and arousal were assessed.

As one would expect, perceived workload was significantly higher for the control condition than for both the CACC with and CACC without merge assist groups. This indicates that controlling speed does increase drivers’ overall perceived workload. One might expect to reveal similar findings while using ACC and traditional cruise control dependent on traffic densities. It is often speculated that those people with lower perceived workload will engage in other activities to increase arousal (based on the Yerkes-Dodson Law).⁽⁴⁾ However, no evidence was found that the different groups engaged in different or more non-driving tasks. Of course, it is reasonable to expect that the demand characteristics of the experimental environment dissuaded participants from participating in other activities given the social stigma associated with distracted driving. It remains possible that drivers may be more likely or willing to engage in non-driving activities under low perceived workload levels.

Despite a perceived increase in workload, no differences in physiological arousal between the different groups were found. The lack of difference between the different groups could be the result of two different causes. The first is that the task participants engaged in was simply too mundane to elicit increases in arousal. In other words, the driving task was so simple and routine that it did not increase arousal levels. The second possibility is that the equipment used to assess physiological arousal was not sensitive enough to capture subtle differences between the different participant groups.

Overall, participants were quite successful in merging into the constant platoon of vehicles. No drivers in the CACC with merge assist group collided with another vehicle while merging. However, as noted previously, if drivers in the CACC with merge assist condition did not override the system or lose control of the vehicle, then it was impossible to collide with another vehicle in the simulation because of the programming of the drive. In total, the control and CACC without merge assist groups experienced 24 collisions, over half (58 percent) of which occurred during the first merge. The improvement over time is likely the result of practice and experience (specifically, the experience of learning that the other vehicles do not respond quite like a real driver might). In the real world, as a driver, one might expect the vehicle behind you on a merging approach to slow and allow a more comfortable gap in order to complete the merge. Several of the participants that experienced a collision commented on the following vehicle not creating a larger gap. For example, one participant said “In my mind, I think he should have let me in,” while another said “Come on, guys…Let me in.” Taken together, it appears that the unanticipated behavior of the other traffic may have contributed more to the collisions that the actual abilities of the driver. With this in mind, it is it important to take other typical behavior that is seen in manual driving into account when designing road technology applications (e.g., other drivers creating gaps or slowing for emergency vehicles).

In sum, it appears that CACC with merge assist may increase the efficiency of drivers’ entering CACC platoons of vehicles. Drivers (with and without merge assistance) were willing to accept speed assistance in the simulated environment. Drivers also willingly entered gaps smaller than 1.1 s, showing that there may be a potential for smaller gaps to be used in a real-world setting.