This is the fourth of four follow-on articles to Do Smart Highways Now—Delay is Not Smart, in which an integrated automated highway system is envisioned. This article will describe what might be called the kinetic component of the system—the movement of physical steel on wheels, with people and cargo inside.
Before automating anything that affects personal safety, all factors must be addressed that humans would attend to when operating something manually. This is not impossible—airliners in flight have been automated for a long time, and arguably for good reason. There’s no reason why similar automated capability could not be applied in cars. Although cars are much more numerous than airliners, they also travel at greatly lower speeds (even those that are speeding) and they are more accessible than aircraft from terrestrial points when underway. At least some of the principles applied to airliner automation could also be carried over into cars.
We start in this by enumerating the factors considered by a human driver when operating a motor vehicle on the road or street. Some of these are learned in driver training, such as how much to turn the steering wheel and at what safe speed when executing a turn. Hands, arms, eyes, head, feet, and ears all play a role that is orchestrated by the brain’s cerebellum in a trained driver. The vehicle flow system must do all of these things as well or better than a human, coordinated as described in the previous article on data and control flow. The “or better” factor is what makes the vehicle flow system worth doing—and in so doing it will reduce traffic congestion and accidents resulting from human error (and lack of big picture traffic awareness).
An experienced human driver will know how to adjust to different pavement conditions such as rain, snow, or ice. The human will remember those times when the car went into a skid upon going over an icy patch—and more importantly what the driver did to recover from the skid and live to tell the story. Similarly, the experienced human driver will remember the proper driving responses to heavy traffic, short stops by the car in front, unsafe maneuvers by adjacent drivers, and excessively fast or slow cars. One of the causes of traffic congestion is that the responses to the same road or traffic conditions are as numerous as the different drivers in the same group of cars that are in close proximity to each other. Each driver’s individual experiences are different—causing a ripple effect of unexpected actions by other drivers that cascade into greatly reduced flow efficiency.
Each individual car is also unique even within vehicles of the same make and model. When was the last time you drove someone else’s car, or got a rental car? Not only were the controls different, but also the steering wheel needed to be turned more or less, the brake pedal needed to be pressed harder or lighter, and the tires had more or less traction than what you were used to in your own car. As a human, your experience from driving different cars in your lifetime provides you with enough background to adjust your driving style to operate that particular car in the most optimal way for you. That is the situation with individual drivers in individual cars—now imagine a centralized distributed vehicle flow system that is driving all of the individual cars as one control process.
Adapting driving to automated control is hardly a far fetched concept. Cars have had onboard computers for a long time, and these have proven their utility. There are several features that are needed at the vehicle level for the vehicle flow system to function as described in the previous articles:
- Each individual car’s control unit (and supporting sensors) would need to be calibrated to the distinctive variances of the car it was installed in, including:
- Steering wheel turn to tire position result
- Brake pedal pressure to kinetic slowing effect, and for different loading (inertia) levels in the car
- Tire traction factor, and for different road conditions—higher performance tires can take harder braking before starting to skid for any given type of pavement or condition
- Drive train acceleration factor, determining the points of downshift (and upshift during deceleration)—this is where control unit communication with the car’s fuel system computer could pay off well
- Physical means (transducers) to convert control unit signals to corresponding action on the car’s operation
- Servo motors in the steering wheel and tire position control components
- Hydraulic, magnetic, or other means of applying pressure to brake pads or disks
- Electric switching (e.g. solenoid or digital) for activation of turn signals and brake lights (for humans in other cars being driven manually)
- Headlights on or off, determined by a combination of (seasonal) time of day and level of ambient light (affected by weather, tunnel, or other light blockage)
- Environmental sensors to provide data on road conditions to guide the control unit on the degree of turn or pressure (such as to compute braking distance)—possibilities include:
- Laser beam from under the car reflected off the pavement to reveal the coarseness of the pavement, which is an indicator of expected traction from any given set of tires (physically similar to the way CD/DVD players work)
- Moisture sensors directly behind one or more of the wheels (to determine the degree of wetness on the pavement)
- Calibrated thermometer (also mounted underneath the car) specifically designed to determine whether the temperature of the pavement the car was instantly on was at or below freezing
All of these data points together would be overwhelming for a human driver. They actually would comprise a dangerous distraction, with no commensurate benefit for either safety or efficiency. But an expert system such as discussed in these articles could handle all of this—it’s actually designed to process and apply all of the data available, both locally in the car’s control unit and globally in the distributed central system. The vehicle flow system may be called an expert system, because it contains the combined experience of engineers and drivers into a practically applied result. (Whether to call it AI is mentioned previously in the data and control flow article—my opinion on the subject is that’s an issue of semantics only and otherwise irrelevant.)
For all of this to become a reality, equipment items would need to be designed and manufactured, and diagnostic devices built for configuring and maintaining them. Then manuals and technical notes would be published to maintenance engineers and technicians at local car dealers and garages. Initial and continuing training and associated certifications would also assure proper maintenance and operation.
Standards (both functional and technical) would need to be established for architects and engineers in all supporting fields to design to. Many standards exist already, such as IPv6, PKI, wireless, networking, computer cloud, open source software libraries, and those in automotive fields. In my opinion, the standards for the vehicle flow system are best curated by established standards bodies, such as IEEE, IETF, ISO, and SAE.
In conclusion, I think it is appropriate to re-iterate the closing statement of the introductory article: No valid excuse exists for not implementing smart highways. The tech is there; just establish the communication and performance standards, and build to them. Do it now.
mcw
—