Machinery and man: interact at your peril
A lot of people think that the distracted driving epidemic began with cell phone talking and texting, navigation screens and GPS units, but problems with distraction began long before this wireless technology became ubiquitous.
Distractions for drivers, inside the car and out, have been multiplying for decades. Digital displays with warning lights and alarms have been standard installations in cars since the most basic fuel gauge and odometer became obsolete.
The human-machine interfaces (HMI) we have today provide an astounding amount of information about the inner workings of the vehicle and the outside world. They have become one of the top selling points for motor vehicle salespeople looking for differentiation from other brands. This is true in every motor vehicle type including such diverse items as construction cranes and farm equipment.
We will explore the benefits and breakthroughs occurring constantly in vehicle HMI, as well as the dangers of events like “too much information”, or TMI.
Drivers have always been taught to briefly glance at signals and gauges in front of them. These manufacturers know that in order to keep safety the paramount concern, they must constantly remind people that the visual features of interfaces are meant to be used in the same way analog dials were used in the past:
• Speed is checked by glancing for a fraction of a second at the speedometer
• Other gauges like fuel and temperature are meant to be checked when stopped, or in the same split-second glance
• The more locations in the dashboard that have displays, the more time it will take to locate the device
• As a general rule, audio voice warnings and acknowledgements are safer that visual ones. We will look at this audio choice in more depth. Its safety possibilities are outstanding.
By its nature, HMIs can help reduce or increase stress in the operator, whether he is operating a machine lathe, farm loader or automobile. (1) In a brief description of the importance of the interface, this report calls for the reduction of “task saturation,” or the intensity in which the operator finds himself in relation to his immediate surroundings. It is certainly true that saturation for a machine operator is different for that of a car driver, but the principle is different: every distraction in a moving vehicle is much more dangerous.
These distractions can be reduced by building input/output systems that use faster, more accurate signal production. “The advantages provided by incorporating HMIs include error reduction, increased system and user efficiency, improved reliability and maintainability, increased user acceptance and user comfort, reduction in training and skill requirements, reduction in physical or mental stress for users, reduction in task saturation, increased economy of production and productivity, etc.,” the report said.
Scenarios of danger and defense
AI driving innovation in autonomous vehicles is far from ready. While development continues at a strong pace, the technology that will allow vehicles to travel by themselves is far from a reality. For many years to come, human drivers will be the primary pilot of the vehicle. They must be in charge continuously because the nature of vehicular travel is that things will go wrong. Accidents due to error, breakdowns, equipment failure and weather will still happen frequently.
Artificial Intelligence (AI) development is streaking forward, and AI standards and structure are trying to keep up with development. The disconnect between hard-charging autonomy development and safety/security standardization is real and widening. Testing involves tightly controlled environments; these are far from real-world situations. The basis for new interfaces must be clarity, brevity and minimal information necessary to complete a task. (2)
This report about designs of the newest HMI indicates that manufacturers are keenly aware of the dichotomy facing information systems designers: the need to differentiate from other manufacturers (especially in consumer passenger vehicles) for dazzling sales purposes vs. the need to minimize glamor and glitz in screen presentations to heighten safety.
“If we overload the driver’s attention, we’re asking for trouble. This is something that happens too often in cars on the road today. Systems are often designed for use in optimal conditions, not accounting for a driver’s already high cognitive load from the very act of driving, conversation with passengers, or pondering what to pick up from the store after work,” the author said.
His own drive home from work in the Boston, Mass. area near Harvard University is fraught with turns and lane changes that challenge everyone. Imagine, he said, the first-time driver attempting the drive with GPS attempting to dictate lane changes much too late. A better HMI must have faster sensors, faster pulse-to-interface time and clear, simple instruction. The author espouses the use of more audio commands and answers – importantly using clearer, briefer language – to battle the tendency drivers have to stare at a screen too long, become confused by visual maps or misunderstand visuals.
Live and die by input-output?
Nearly every problem with distraction occurs while the vehicle is moving. Stalled or poorly visible vehicles do not cause accidents related to driver distraction, but distraction in moving vehicles causes many collisions with stopped vehicles. For this reason, the faster the information displays on the screen, or is reported vocally, the less distracted time the driver will encounter.
Part of the solution to faster reaction of the display unit is more sensitive sensors and input screens. The main controversy about sensing today is the driving environment coverage of autonomous vehicles, but systems sensors can play a large role internally for warnings and other information sent to the driver. A more precise, faster touchscreen will also reduce distraction time. (3)
Also, the ability to input more pre-settings will help drivers focus more on the road and less on the interior of the car. More programming for automatic settings, like climate control and lighting, will ensure more safety through less distraction while moving. Presetting are easy to change and rarely need changing while the car is in motion. Pre-settings help prevent the “TMI syndrome” that distracts drivers.
Too much information
The problem of too much information contained in either input or output interfaces, particularly in GPS-enabled search systems, can be tackled in several ways. We have already touched on audio interfaces that can allow drivers to keep their eyes on the road. This system can help by allowing input and output using voice commands.
Companies are now researching better language recognition software to bring normal conversation recognition to mainstream automobile interfaces. Nuance Communications, Inc. (sponsors of reports (2) and (3) noted above) has researched speech recognition for many years and created Dragon® software that enables users to “train” the program vocally to recognize their voices. This company and others hope to build vocal recognition into mainstream automotive interfaces so each driver can teach the software over time to learn nuances of each voice.
Whether voices are fast or slow, or interrupted or changed in volume, smart vocal software can recognize words and context. The result would enable users to interact with systems through voice only, leaving most or all hand-controlled commands to be done when the vehicle is stopped. Voice control is fundamental for safety when the vehicle is moving.
The most intricate audio usage today is GPS location over-voiced software. Our phones and cars have been telling us where to go for years, but the technology can be improved. As always, there are complex instructions and regulations for interfaces, including performance and reliability. Asking an audio interface for directions is less stressful that asking for emergency services in an emergency, but the equipment is expected to perform as flawlessly as possible in both cases.
Putting voice to concerns
In 2016, the National Highway Traffic Safety Administration reported on evaluation practices for voice-activated and voice-recognized interfaces in use at the time, and evaluation requirements for the future (4). The report attempted to qualify and quantify errors inherent in audio response systems as compared with touch systems. It also laid out guidelines for testing vocal recognition reliability.
The system tests were designed to reinforce the idea that vocal/audio systems allow better driver safety in which he keeps his eyes on the road for more consistently. Tests also were gauged to show where “speech recognition errors, complex interactions, and response delays might all draw drivers’ attention away from the road.”
As a means of checks and balances in testing systems, the NHTSA used three studies independently. “The three studies included an on-road contextual interview study, a driving simulator study, and a laboratory-based collision detection task study. The CDT is an alternative driving surrogate task that requires the study participant to monitor a dynamic animation on a computer screen to identify spheres that appear to be on a collision course with the observer.”
The report concluded that more complex tasks take longer to process, whether visually or audibly, and that in most cases audible references are preferable to visuals. This reinforced the idea that the less drivers look away from the road ahead the better. Both interface types need more speed and accuracy to become reliable throughout the driving realm.
Finally, we take a look at further automating systems for safety and reliability. This will be far more critical in an autonomously operating vehicle in the future.
The main components of every system must not only be designed to simplify the activities of the driver, whether he is an active driving role or passive monitor in an autonomous vehicle, but also build driver trust in the systems. (5) As the command/response reliability builds, so does trust in the system.
This report looks at driver behavior in many situations but also interprets his trust level combines with experience. “Driver-initiated automation” carries conscious and subconscious trust factors about the reliability if the machinery. Also, some factors that might not be on the top of designers’ minds: “… there are still concerns that after prolonged exposure to highly automated driving, driver desensitization may occur resulting in a lack of task engagement. Questions remain over what the driver will actually do with this increased attentional capacity. It seems reasonable to suggest that drivers may be more inclined to engage in secondary tasks and if this happens, manual override in unanticipated and unexpected events will be difficult to manage …”
AI struggling with human intelligence in an anti-fatigue battle is one of the psychological problems inherent in the secondary role a driver/monitor will face. While this report centers on driverless vehicles, it is well within the purview of distracted, bored or agitated driving that happens daily on roads everywhere. Humans are the driving force of intelligence for all these systems, yet fatigue and emotional distress lead to the very errors the systems are trying to avoid.