Many car enthusiasts say they expect to see self-driving vehicles on roads soon. That’s understandable. Eager automakers and tech companies are in the news often about autonomous vehicle goals. Last year, Detroit and Silicon Valley hoped to see self-driving taxis in service sometime in 2019. However, it’ll likely to be years before we see widespread use of autonomous cars.

Jumping the gun technically

The development of self-driving cars requires collaboration between automakers and tech companies. Optimism for this segment has been high for years. The problem is that companies have underestimated the challenges of producing safe self-driving vehicles.

The process is difficult, expensive, and time-consuming. “We overestimated the arrival of autonomous vehicles,” Jim Hackett, Ford’s CEO said at the Detroit Economic Club early in 2019. What is the biggest cause for the delay? Not surprisingly, it’s human behavior.

Ford and Volkswagen announced they will be collaborating on autonomous vehicle solutions. In 2021, they’ll work with Argo AI, a Pittsburgh startup, for ride-sharing services in select urban areas. Argo’s CEO, Bryan Salesky, said autonomous vehicles were “way in the future.”

He explained that his company, along with others working toward the same goal, have developed as much as 80% of the technology needed to launch self-driving cars. It’s the other 20% that’s delaying things. What remains is the technology needed to consistently predict what cyclists, other drivers, and pedestrians will do next. The process isn’t simple.

Last year, many executives believed most technical issues would be resolved. They thought people would be traveling in driverless vehicles in many cities by now. General Motors and Waymo, owned by the same parent company as Google, claimed they’d have driverless cars in service by the end of 2019.

Investors were eager to jump on the self-driving vehicle bandwagon. Amazon hopes the segment can help it get orders to shoppers. Honda invested in General Motor’s Cruise. Most thought between existing artificial intelligence and sophisticated sensors, autonomous vehicle development would be straightforward and easy.

Safety above all else

Last year, an Uber test car struck and killed a pedestrian last year. A woman in Tempe, Arizona was walking a bicycle across the street. According to Tempe Police, a safety driver was in the vehicle but she was watching a TV show on her phone. It’s believed to be the first pedestrian casualty involving self-driving vehicles.

After the tragic accident, interested parties focused on human safety. Three Tesla drivers also died in crashes with the automaker’s Autopilot driver-assistance system in use. In those U.S. cases, both the system and drivers failed to properly identify and react to dangers they encountered.

Limitless situations — driving on the wrong side of the road, making illegal turns, or running red lights — must be accounted for when developing self-driving cars. Pedestrians jaywalking or crossing the street during a green light for cars.

Another challenge is teaching autonomous cars “micro maneuvers.” The vehicle’s programming must keep it from tailgating other vehicles. They should be able to identify when another car abruptly moves into their path. Now the technology exists that can keep a car from striking anything. But such vehicles would apply the brakes often and be over-cautious.

Companies like General Motors and Waymo are still discussing the release of driverless-car fleets. They are evasive concerning a timeline. Mary Barra, General Motor’s chief executive, said their Cruise division was progressing “at a very aggressive pace” in June. She didn’t say when commercial operations would start.

Not there yet

Along with Uber, General Motors, Ford, Waymo, and Volkswagen, many other companies are hard at work on autonomous vehicles. The California Department of Motor Vehicles has approved 52 companies, including Apple, BMW, Honda, Intel, and Nissan, to test self-driving vehicles. 

Tesla is pretty much alone now in predicting the widespread use of autonomous cars by next year. CEO Elon Musk said his company would have up to a million self-driving “robo taxis” in service by the end of 2020.

Tesla is confident its Autopilot driver-assistance system, as well as the data harvested from Tesla cars currently on the road, will allow them to start offering driverless cars. Experts have their doubts that Tesla will pull off that ambitious goal.

Some smaller services are using driver-less technology on a very small, limited scale. A start-up in Boston, Optimus Ride, plans to offer driverless shuttles that will travel at 25 mph or less at the Brooklyn Navy Yard.

The Rhode Island Department of Transportation supports a project for the May Mobility shuttle service in Providence. Since May, they’ve run six shuttles between Olneyville, a growing nearby neighborhood, and the Providence train station. May Mobility also has a partnership with the city of Grand Rapids, Michigan and hopes to take its service there.

These small services are not an indication, however, that driverless cars are right around the corner. Some in the industry say it could be as much as a decade before we could see widely used autonomous vehicles.