Cooperative Adaptive Cruise Control: Human Factors Analysis

IMPLEMENTATION ISSUES

This analysis assumes that the technological requirements for the CACC concept are viable and function as expected. However, technical validity does not necessarily translate into successful operation and implementation. There are a host of human-factors issues that may come into play and affect if and how CACC technology is utilized.

Automation

Several benefits have been identified for the application of automation in the driving environment. In addition to the throughput and environmental benefits of the CACC concept, other driving-related automation has been touted to improve performance and reduce driver stress, error, and workload.(19) As a bonus to car manufacturers, new automation may increase car purchases by people that desire the latest technologies.

The benefits of the CACC concept stem from automated throttle and braking to permit vehicles to follow more closely, removing or reducing human interaction from the longitudinal control. These improvements are attained by addressing key mechanisms in the braking process, which can be broken down into five key components, as follows:⁽²⁰⁾

• Perception of an event or lead-vehicle slowing, including brake light onset.
• Processing of information to interpret the event.
• Decision to take action.
• Selection of the appropriate response.
• Initiation of the response.

The first four components are cognitive and subject to numerous delays, depending on the environment, SA, workload, and individual differences of the driver, to name a few. It has been demonstrated that by using automation, vehicles are able to adjust to speed changes more quickly than solely by manual human reaction. Numerous studies have measured human brake response time (BRT), the time from event onset (usually brake lights of the lead vehicle) to initial pressure on the vehicle's brake pedal, under various conditions. When research study participants are perfectly attentive to a simulated driving response task, average human BRT can be as little as 0.47 s.⁽⁵⁾ Age, gender, and training play major factors, as well. Young athletes, who tend to have higher than average hand-eye coordination, have been shown to have BRTs in the 0.51 - 0.55-s range in real traffic.⁽²¹⁾ Younger drivers, in general, have quicker BRTs, and reaction times tend to increase with age. Additionally, male drivers typically have faster BRTs than female drivers.⁽²²⁾

When braking is aided or controlled by an automated system, the cognitive delays that humans inherently express are all but eliminated, allowing initial brake application to occur remarkably faster, in less than 0.1 s.⁽²³⁾ It is important to note, however, that braking capabilities with CACC are primarily geared toward maintaining a set time gap by utilizing specific acceleration and deceleration models to provide a comfortable ride for the vehicle's occupants. Emergency stopping maneuvers require significant brake force and may or may not be part of an implemented CACC system. Other technologies, such as collision avoidance systems, may complement CACC to provide automation for emergency situations. In any event, it is critical that drivers understand the limits of automation and utilize it as intended.

Willingness to Utilize Automation

At the heart of the CACC concept is the ability to increase traffic throughput by vehicles traveling closer together. Even if the technology is sound, it will only be successful if drivers are willing to travel more closely.

Most official guidelines from highway-safety organizations suggest a 1 - 2-s gap.⁽⁵⁾ Research studies looking at participant-specified comfortable time gaps in manual driving have generally supported this guidance. In one study, 95 percent of the participants followed the lead vehicle at a gap of 1.68 s or less and had an overall average of 0.98 s, which supports findings from previous comfort gap studies.^(5,24,25) Similarly, a study having participants either fall back to or approach a comfortable following gap revealed an average time gap of 1.1 s.⁽²⁶⁾ A study looking at how time gaps affect perception of risk, difficulty, effort, and comfort found that all measures were rated low until the time gap dropped below 2 s and continued to climb as the gap decreased.⁽²⁷⁾ In the study, the general range of time gaps selected was 1.67 - 1.78 s. Two naturalistic studies also supported these time gaps for manual driving, with averages of 1.64 and 1.6 s.^(28,29)

Automation of time gaps using ACC or CACC provides the driver with a few preset time gaps to choose from. A recent quasi-naturalistic study comparing gap acceptance between manual, ACC, and CACC driving found that drivers were willing to drive at closer time gaps when using automation.⁽²⁸⁾ Whereas the average manual time gap was 1.64 s, the most commonly utilized ACC gap setting was 1.1 s (selected in 50.4 percent of engagement time), which was the shortest of the available time gaps (1.1, 1.6, and 2.2 s). Options for the CACC system were 0.6, 0.7, 0.9, and 1.1 s, and the shortest time gap was selected in 55 percent of the system's engagement time.

Somewhat paradoxically, research showed that while drivers rated their comfort with fully automated driving very high when they were the lead vehicle, comfort dropped to a negative rating for over 70 percent of the participants when a merged vehicle became the lead.⁽³⁰⁾ Similarly, an ACC study that involved naturalistic driving in three European countries indicated that such systems are generally viewed as a comfort system and utilized less frequently when traffic is dense, which is when the CACC concept would have its biggest potential impact.⁽³¹⁾ So, although research has shown closer following distances to be acceptable, some reluctance to rely on automation or timely utilization exists and could affect the actual utilization of CACC technology.

Additionally, some drivers may choose not to utilize automation due to its restrictions. A study looking at driving behavior with ACC categorized the participants based on their responses to a driving style questionnaire.⁽³²⁾ Drivers delegated to the "speed" group, where driving fast appeared to be a chief priority, identified the ACC system as uncomfortable and not useful.⁽³³⁾ So, even though the CACC system may technically provide a more efficient means of transportation, some drivers prefer to manually manage speed and maneuvers. Unfortunately, the driving behavior of this group is likely a promoter of instability in traffic flow and would benefit most from the CACC system.

Application of Automation

Though automation is usually proposed to "supplant human activity," it typically just changes the nature of the human role, which may have unanticipated or unintended consequences.⁽³⁴⁾ Automation may be able to perform at a higher level than humans, as is the case with BRT, but humans are usually left in charge to monitor the system, leading to a variety of potential issues. As automation allows a system to perform better than if it were manually controlled by a human, system failures may put the human in his least capable situation.⁽³⁵⁾ In the case of CACC, drivers would be following a lead vehicle at a much shorter time gap than they may be able to accommodate in the event of a CACC system failure.

In monitoring roles, humans have been shown to perform with less than stellar degrees of success. Studies have demonstrated that drivers perform worse when reacting to automation failures than to critical events under manual control and that performance diminishes as levels of automation increase.⁽³⁶⁾ Automation use in other industries, such as aviation and maritime, have provided many instances in which monitoring failures have resulted in untimely, inappropriate, and even non-existent human responses. (See references 37 - 40.) Many of these issues pertain to how the automation was understood and utilized.

Trust and Reliance

In order for automation to be utilized, a certain level of trust must exist between the user and the technology. Trust evolves over time in complex individual, cultural, societal, and organizational contexts and is usually based on a technology's ability to achieve a particular goal.⁽⁴¹⁾ Automation utilization requires a user to be vulnerable to the automation's actions with the expectation that it will be successful in helping the user achieve a specific goal, and correct utilization of automation requires that the correct level of trust be placed on it.⁽⁴²⁾ Incorrect levels of trust result in three possible outcomes, as follows:⁽³⁴⁾

• Misuse: Users violate critical assumptions and rely on the automation inappropriately.
• Disuse: Users reject the automation's capabilities and do not utilize the automation.
• Abuse: Designers introduce an inappropriate application of automation.

How someone determines their level of trust in automation depends on an accurate understanding of the purpose, operation, and historical performance of the automation.⁽⁴³⁾ Unfortunately, users do not always make the correct assessments of these components and often use or rely on automation inappropriately.

A primary and understandable goal for most automation is high reliability. A system with a high failure rate, after all, is not likely to achieve much long-term success. While high reliability fosters trust and likely results in accomplished goals, it also promotes an undesirable side effect. Several studies have shown a complacency effect for highly reliable systems, where users tend to over-rely on the automation, using the automation beyond its intended scope or failing to adequately monitor for malfunctions. Novice users, who may have never experienced a system failure or have only been exposed to automated functionality, may not be able to recognize a malfunction or adequately regain system control when necessary. Surveillance task studies have revealed troubleshooting complacency for participants that had not experienced system failures during training sessions.⁽⁴⁴⁾ Reliance bias increased with those that had practiced system failures as system reliability began to increase. Monitoring performance has been better in variable reliability trials than in totally reliable trials, indicating over-reliance on highly reliable automation.⁽⁴⁵⁾

Experienced users are equally prone to over-reliance and complacency when system reliability is high. Pilots have been shown to rely on automation in situations well beyond the system's intended use and to ignore conflicting evidence of expected and actual automation performance.^(19,38) In 1995, a passenger ship ran aground near Nantucket, MA, when the crew blindly relied on a failing navigation system, ignoring numerous position information system displays clearly indicating the ship was drifting off course.⁽⁴⁰⁾

Reliability is not the only factor involved in system trust, however. An interesting study on ACC looked at the relationship between participants' mental model of how the ACC system functioned and their level of trust with the technology.⁽⁴⁶⁾ Over 10 consecutive days, participants interacted with ACC technology using a driving simulator. At the end of each day, participants provided a rating of trust and a graphic representation of how they understood the ACC technology to operate. For the groups in which the ACC system malfunctioned 50 or 100 percent of the time, mental model representations changed each day and level of trust with the system never increased. For the group in which the ACC system functioned flawlessly 100 percent of the time, level of trust ratings did not climb until the fourth day, which coincided with the day in which the mental model of the system became fixed. This study demonstrated that having faith in one's understanding of how a system works improves the level of trust in technology, possibly even more than reliability itself. Interestingly, participants' graphic representations, while unchanged after 4 days, never matched the actual system model. This indicates that drivers may end up fully trusting a system by incorrectly believing they understand how it functions, which can lead to inappropriate usage of automation.

Improper trust in automation has been shown in several ACC-related research studies. In a study comparing automated and manual driving, the majority of participants in the automated driving scenarios braked in emergency situations only after a collision alert sounded, indicating they were waiting on the automation to react rather than maintaining an active role.⁽⁴⁷⁾ Comparing manual driving with the use of CCC and ACC in fog conditions, another study found that average speeds were significantly higher with ACC use, signifying that drivers were relying on ACC to slow the car when necessary, even in reduced visibility.⁽⁴⁸⁾ (Fog can also have a negative impact on the functioning of the ACC system.) That same study also showed that speeds approaching curves in fog conditions were reduced much later in both CCC and ACC scenarios, again suggesting that drivers may utilize automation at inappropriate times, either due to over-reliance, misguided trust, or misunderstanding of the automation's intent or capabilities.

Carryover Effects

One effect of automation on driving behavior that has not been studied much is the impact it has on driving performance after returning to manual control. In the instance of CACC, drivers may become accustomed to driving at very close time gaps. If a driver were to continue at such a gap after switching to manual mode, this could create an extremely dangerous scenario.

An extensive set of experiments in a study on fully automated driving included two scenarios that evaluated carryover effects when switching back to manual driving.⁽³⁰⁾ The first scenario reflected mixed results, where lane-keeping behavior was better, speed control was worse, and selected time gaps were unaffected when the driver switched back to manual control. However, when the automated driving period was extended before switching back to manual control (4 min in one scenario versus four consecutive trials of nearly 30 mi), significantly smaller time gaps were selected in manual driving. As drivers become more accustomed to CACC technology, this potential carryover effect stands to be a legitimate concern.

Workload

Although there are several theories on what attention is and how people allocate it, all ultimately concur that there are limits to how much information a person can attend to at one time. This limit may vary based on the specific task, a person's level of arousal or experience, and the ultimate goal, but at any single point in time, there is an upper limit on what can be processed.⁽⁴⁹⁾ Workload is the overall level of attention demand a task (or group of tasks) presents. As demand for or complexity of one task increases, one's overall workload increases and the ability to attend to new information decreases. The more experience someone has with a task, however, the less demanding the task becomes and the less impact it has on the person's workload levels. Therefore, novice driver workload levels are often maxed out with typical vehicle control tasks, which leaves them with little capacity for other driving-related tasks, such as hazard identification and prediction.⁽¹⁹⁾

As previously stated, one touted benefit of using automation in vehicles is to reduce driver stress and workload.⁽¹⁹⁾ Automation removes the need for a driver to actively perform a specific action, and the driver theoretically has more cognitive ability available for other actions. Several studies have shown that technologies such as ACC reduce workload levels, and it is reasonable to believe that CACC should realize similar gains.^(50,51)

Although reducing workload is usually a positive result, reductions below a certain level can have a negative effect. Human performance is optimal when workload levels are in between extremes, as professed by the Yerkes-Dodson Law.⁽⁴⁹⁾ When arousal levels are too low, humans tend to perform below their abilities. As arousal levels increase, so does performance, up to a point at which a task begins to overwhelm human capabilities and performance begins to suffer. This trend creates an inverted U shape when graphed - a positive slope up to some threshold, followed by a negative slope when arousal levels surpass competencies. Reducing a driver's workload level frees up attentional resources but can reduce arousal to the point at which performance suffers. A study comparing manual and fully automated driving found that in automated conditions there was a significant decrease in heart rate and percent road center gazes, which pertain to a region surrounding the most common fixation points.⁽⁵²⁾

These physiological changes can translate into negative consequences while driving. In a study in which workload levels varied, participants reported an increase in mind-wandering during low workload scenarios and demonstrated a reduction in horizontal gaze dispersion and side mirror checks.⁽⁵³⁾ Previous research has indicated that when the primary task does not require executive control in the brain, it permits one's focus to switch to internal information processing (i.e., mind-wandering).⁽⁵⁴⁾ As this underload occurs, delayed reactionary performance can occur, which could have catastrophic consequences when traveling at short CACC time gaps.^(36,47,51)

Automation-reduced workload does free up more attention capacity for a driver. This can be a huge benefit when the driver uses this available capacity for driving-related tasks, such as scanning for hazards or predicting future states of the driving environment. Unfortunately, increased attention capacity does not necessarily translate to increased driving-related performance. Drivers may also be encouraged to attend to non-driving-related tasks.

Distraction and Situation Awareness

With a portion of the driving task aided by automation, the driver has the ability to put additional attention resources toward improving surveillance performance or other driving-related tasks. However, numerous studies regarding driving automation, including ACC, have demonstrated that this spare capacity is often used to engage in non-driving-related secondary tasks. Radio interaction and DVD player usage, number and duration of off-road glances, and other secondary tasks all increased under some form of automated driving.^(52,55,56) Additionally, tests on such secondary tasks showed that performance on these tasks improved under automated driving, which demonstrates the additional attention allocated to them.^(57,58) Apparently, the more driving automation involved, the more drivers are willing to rely on automation to permit them to perform non-driving related tasks.

Increased secondary task engagement has a direct impact on a driver's SA, the driver's perception of various elements in the driving environment, comprehension of their meaning, and prediction of their status in the near future.⁽⁵⁹⁾ These three components of SA can be viewed as the operational, tactical, and strategic levels of driving, which incorporate navigational knowledge, environment and interaction knowledge, spatial orientation, and various vehicle statuses.^(60,61) Any increase in non-driving-related secondary tasks decreases these SA knowledge sets and, therefore, negatively impacts driving performance. Emergency situations, such as an unexpected conflict or automation system malfunction, require quick reactions, which are based on an appropriate SA level.

Several studies on BRT with automation clearly demonstrate the potentially disastrous effects distraction and reduced SA can promote. Even when participants were expecting the braking event or had ample information to anticipate the need to brake, drivers utilizing ACC had much higher BRTs than those manually controlling the vehicle.^(20,47,57) The braking performance itself also demonstrated reduced SA, as deceleration rates with ACC were twice that of CCC and significantly less safe with ACC when compared to manual driving.^(55,57) Similarly, participants in automated driving scenarios in other research studies only applied the brakes after a collision alert sounded, significantly reducing the minimum time to contact.⁽⁴⁷⁾ They also exhibited the worst performance when trying to regain driving control from the automated system.⁽⁶²⁾

Participants in early research on ACC provided feedback indicating they liked being able to feel the ACC system deceleration, as it made them aware the system was reacting to some conflict.⁽⁵⁵⁾ This indicates that rather than maintaining the necessary SA for conflict identification, drivers were relying on automation to alert them when it was necessary to surveil and take action. With closer following gaps, CACC usage may not permit adequate time for a driver to recover from failed SA maintenance.

Several behavioral theories may explain willingness to undertake secondary tasks. Risk Homeostasis Theory speculates that people seek to maintain a certain level of risk. As an environment becomes safer, riskier behavior may transpire; as an environment exceeds one's acceptable risk level, fewer risks will be taken.⁽⁶³⁾ If automation is perceived to make the driving environment safer, sensation seekers may be more willing to engage in risky behavior, such as non-driving-related secondary tasks.⁽⁶⁴⁾ Research on such individuals has shown that they perform better at secondary tasks, take longer to respond to lead vehicle brake lights, initiate more unsafe braking events, and demonstrate worse lane position variability when driving with ACC engaged than when driving manually.⁽⁵⁷⁾

While CACC may enable drivers to travel more closely to a lead vehicle and provide additional traffic-related information to the driver, the automation may have unintended consequences that reduce a driver's awareness of the surrounding environment.

Driving Behavior

In addition to being required for accurate microsimulation modeling, an understanding of general driver behavior is necessary to determine areas in which automation such as CACC may pose risks. Driving behavior studies typically involve areas such as lane-changing, car-following, turning, acceleration, and deceleration; of particular concern for CACC technology are lane-changing and car-following. Although these actions are directly measurable, the motivating forces behind them are more difficult to ascertain and may be prone to human error.

Lane-Changing

Though the CACC concept would eventually apply to all travel lanes, the throughput benefits would be greatest when drivers resist changing lanes and remain in a platoon as much as possible. Not only is a lane change a generally risky maneuver, often involving quick decisions and issues with blind spots, but it can be very disruptive to traffic stability. Several studies have been done to determine why and when drivers change lanes. The results provide conflicting predictions for CACC utilization.

In one driving simulation study, participants displayed a strong tendency to pass a lead vehicle regardless of the lead vehicle's speed.⁽⁶⁵⁾ As expected, when the lead vehicle was traveling slower than the participant, the drivers passed in almost every instance. When the lead vehicle was traveling at the same speed as the participant, the participant passed in 66 percent of encounters. Surprisingly, even when the lead vehicle was traveling faster than the participant's average speed, the participants passed roughly 50 percent of the time. What was not clear from the study, however, was why the lane changes were performed. Passing a slower vehicle is obvious, but were the other lane changes due to driver aggression or an increased perception of risk, effort, or workload? The authors assert that one potential cause for passing vehicles moving faster than a driver's average speed is the variability in speed a driver may exhibit. If a lead vehicle is traveling at a speed within a driver's speed variability, the driver is likely to be traveling faster than the lead vehicle at some point and will be more likely to pass. If this is a key factor in lane-changing behavior, CACC technology could provide a big benefit by reducing variability.

Research into how drivers perceive the speed of vehicles in adjacent lanes also provides insight. In comparisons of simulated traffic in two lanes and in actual field-recorded traffic observations, one lane was generally perceived to be traveling faster even though the overall average speed was identical or slower in the selected lane.⁽⁶⁶⁾ The general thinking behind this misperception is that vehicles spread out when traveling faster and bunch together when traveling slowly. This makes passing epochs very short (i.e., a fast-moving driver passes many cars in a short period of time) and overtaken epochs very long (i.e., it takes much longer for the same number of vehicles to pass a slow-moving driver). Drivers do not tend to integrate the frequency and duration of these epochs and wrongfully believe the "good" (passing) frequency should equal the "bad" (overtaken) in order for average speeds to be equal. Any difference makes the driver believe he is in a faster or slower lane. Further accentuating this illusion are superficial characteristics, such as a powerful sounding engine or squealing brakes, and the frequency with which a driver tailgates or glances at adjacent lanes.⁽⁶⁶⁾ Misperceptions such as these may cause drivers in a higher average speed CACC lane to believe that adjacent non-CACC lanes may be faster, breaking down trust in and usage of the technology.

Untimely lane changes may also be explained by common decisionmaking principles. Utility theory roughly prescribes that decisionmaking is heavily influenced by end goals and final outcomes - people will make decisions based on whatever is likely to provide the best result. However, numerous conflicting examples have given rise to prospect theory, where decisions are based on potential losses and gains rather than true end results.⁽⁶⁷⁾ People tend to be risk-averse when posed with potential gains and risk-seeking when posed with potential losses. Therefore, when a driver believes he is losing ground to overtaking vehicles, he may be more likely to change lanes in order to reduce the potential losses.

One study showed willingness for drivers to reduce lane changes under automated control, but the motivation was likely not a positive indicator for the CACC concept. In comparison to manual driving in simulated heavy traffic scenarios, researchers found that under automated control (longitude and latitude), drivers tended to remain in a lane even when the adjacent lane was moving faster.⁽⁵²⁾ However, these drivers were also much more likely to engage in visually demanding secondary tasks, which may have precluded them from noticing that adjacent traffic was moving faster and indicates reduced attention to the driving task.

Related to lane changes are issues with joining and exiting a CACC platoon. With V2V communication focused on keeping vehicles at very small gaps, it becomes very difficult for a CACC-equipped vehicle to join an existing platoon at any place other than the beginning or end. Similarly, a vehicle attempting to exit a platoon will likely need to adjust its speed to prepare to merge to an adjacent lane, possibly upsetting the stability of the platoon. A microsimulation looking specifically at a merge scenario due to a lane drop demonstrated how platoons negatively impact the merging process.⁽¹¹⁾ The researchers' suggestions for future research included limiting the length of platoons, infrastructure-based gap-lengthening signals when a downstream bottleneck exists (e.g., construction, lane drop, accident), or additional CACC capabilities to communicate and facilitate lateral merge needs (e.g., turn signal activation by a CACC-engaged vehicle could increase gaps in the applicable part of a platoon).

Car-Following

Car-following is the primary component of the CACC system. Therefore, it is critical to understand how humans behave when behind a lead vehicle and what affects that behavior. A key component of car-following is the time or distance gap behind a lead vehicle. As previously stated, differences in time-gap selection are affected by numerous variables, including age, gender, and weather. Additionally, studies concerned with the use of automation (CCC, ACC, and CACC) have shown drivers are generally willing to travel at shorter gaps than under manual control. What is not clear, though, is if comfort with shorter gaps is based on an accurate interpretation of the environment by the driver.

When estimating following gaps (time and distance), participants in one study were relatively accurate concerning distance but very bad when judging time.⁽⁵⁾ The average time-gap estimate in the study was 2.1 s, but 93 percent of the actual gaps were less than 1 s. Studies have also shown that drivers follow larger vehicles more closely even though visibility is reduced.^(29,68,69) One possible reason for this is the belief that because larger vehicles take longer to brake, the following driver has more time to react and brake. Research shows that, although braking time is only 8.5 percent longer for larger vehicles, participants follow 14 percent closer.⁽⁷⁰⁾ When participants were asked to order vehicles in terms of following distances from short to long, they ordered them passenger car, pick-up truck, bus, and tractor-trailer; however, during driving simulations, researchers did not find any gap differences among the three larger vehicles, indicating that vehicle size may actually change gap perceptions for the driver.⁽⁷¹⁾ Similarly, a study found a discrepancy in gap perception by asking participants to follow at a comfortable gap but varying the starting gap from either extremely close to distant.⁽²⁶⁾ When participants started far away from the lead vehicle, they closed to an average of a 1.46-s gap; when starting very close to the lead vehicle, they fell back to only a 0.7-s gap.

In addition to perceptual issues, drivers suffer from poor judgment that can affect how and when CACC technology is utilized. Humans are often poorly aware of their skills, typically overly optimistic and miscalibrated.⁽⁷²⁾ Drivers are prone to overestimating their own performance and underestimating that of others. Asked to rate their performances after a driving simulator study, drivers believed they performed better in automated driving than manual even though anticipation was better, braking was initiated earlier, and minimum time to contact was higher in manual driving.⁽⁴⁷⁾ In driving distraction research regarding cell phone conversations, even after witnessing other cell phone users driving erratically, half of the participants indicated they did not find driving while on a cell phone any more difficult; study results indicated that all participants demonstrated performance decrements.⁽⁷³⁾ Similarly, another study asked participants to estimate the effect cell phone conversation would have on their own driving performance as well as ranking themselves compared to the average U.S. driver on various skill and safety items.⁽⁷⁴⁾ In most instances, drivers did significantly worse than they estimated, and in some instances, those that estimated the smallest effect of the cell phone conversations were actually the worst in the group.

In research on acceptance of short gaps, participants were asked to predict the likelihood of accidents for themselves and others at various following gaps. Participants consistently rated others as being more likely to have an accident.⁽²⁷⁾ Although this study did not test accident rates, it reflects the overconfidence drivers tend to exhibit regarding their own skills and the underestimation of others' skills. This may prove to be an issue when CACC drivers are followed closely by other CACC vehicles; if drivers are not as confident in the following driver's abilities, stress levels may increase.

Poor judgment, as demonstrated by the above research, may have important implications for how safely CACC could be used. Improper confidence in one's abilities, when paired with the effects of reduced workload, may exacerbate the tendency to engage in non-driving-related tasks. Furthermore, previous research on trust and reliance suggests that trust in automation can itself increase the likelihood of a negative result, where a driver may not properly adjust the use of the automation given signs or history of malfunctions.⁽⁵⁷⁾

Issues summary

Table 1 provides a summary of the human-factors issues described in this chapter and the related research utilized for the analysis.

Table 1. Overview of human-factors issues.

BRT = Brake response time.
CACC = Cooperative Adaptive Cruise Control.
SA = Situational awareness.