Self-driving cars are all about information technology, artificial intelligence and the Internet of Things. In fact, they utilize some of the most complex, and still unperfected hardware and software in the lab. When the remaining issues in autonomous vehicles are solved, however, we will not only experience (promised) carefree driving but also advances in robotics and other human-computer interactions.
In the meantime, solving scientific and engineering obstacles aren’t the only roadblocks for self-driving cars. The driving public is deeply skeptical of the technology. A Gallup poll taken earlier this year showed that 54 percent of Americans say they would be uncomfortable as a passenger in a self-driving car. Even more are worried about sharing the road with self-drgalliving trucks.
On the other hand, Gallup points out, a poll taken in 2000 showed that 23 percent of U.S. adults said they would never use a cell phone.
Development of self-driving cars is following two tracks. Alphabet, the Google spin-off, uses lidar (a laser measurement system) to detect objects around the car, as well as radar. Tesla uses a system of cameras, radar and ultrasonics.
A web of sensors in both systems connects the detection equipment to the gas pedal and brakes. And both also rely on software algorithms to recognize objects and conditions that affect how the car should move. The algorithms, so far, are the most difficult solution to address.
That is in part because while human drivers are error-prone and often erratic, they expect a self-driving car to be blemish-free. The software that owns the market will have to be, if not perfect, able to withstand the scrutiny of a Six Sigma black belt.
“By far the most complex part of self-driving cars, the decision-making of the algorithms, must be able to handle a multitude of simple and complex driving situations flawlessly,” Alan Amici, a vice president of automotive engineering at TE Connectivity told The Franklin Institute. “The software used to implement these algorithms must be robust and fault-tolerant.”
Of course, many car companies already employ elements that will one day go into a fully self-driving car. Today’s high-end cars can parallel park, hit the brakes to avoid collisions and warn drivers when they inadvertently leave the driving lane.
A majority of U.S. states have passed at least minimal regulations to encourage the testing of autonomous vehicles, with Connecticut the latest to allow municipalities to submit applications to participate in a Fully Autonomous Vehicle Testing Pilot Program.
Private and public research institutions are gearing up as well. One example is the partnership between Florida Polytechnic University and the state’s Turnpike Enterprise, which are developing a 400-acre site between Tampa and Orlando to test transportation technology. The enterprise will include a simulated urban streetscape to test interactions between self-driving vehicles, pedestrians and bicycles.
Automakers predict that semi-autonomous cars that continue to require an attentive human in the driver’s seat will be on the market by 2020.
In that near-term, computer-assisted driving will probably evolve with trucks and buses on prescribed routes in confined areas, such as college and industrial campuses. That will be followed by highway driving, which is somewhat more predictable than travel on secondary streets.
Fully autonomous vehicles are at least another decade away, as McKinsey & Co. reported.
“Given current development trends, fully autonomous vehicles won’t be available in the next ten years,” McKinsey wrote. “The main stumbling block is the development of the required software. While hardware innovations will deliver the required computational power, and prices (especially for sensors) appear likely to go on falling, software will remain a critical bottleneck.”
Until then, we will have to console ourselves with the media obsession that produces daily articles on autonomous vehicles. And, if we’re lucky, maybe a ride-along in one of the more-advanced prototypes.