Back to the list

It’s 2023. Where are self-driving cars?

As technology advances, the concept of self-driving cars is becoming more than just a dream. With companies such as Waymo, Cruise, and Uber investing heavily in autonomous vehicle technology, it is clear that the era of self-driving cars is upon us.

Self-driving cars were once promised to be the future of transportation, but the reality is that fully autonomous vehicles are still not available for public use. Despite the efforts of major companies in the tech and automotive industries, such as General Motors, Google's Waymo, Toyota, and Honda, self-driving cars are limited to special trial programs.While you can buy cars that come equipped with advanced driver assistance features like automatic braking and lane-keeping assistance, the engineering challenges of making self-driving cars work reliably and safely have caused many of the ambitious predictions made in the past to be rolled back.Here are five common questions you may have about self-driving cars and why the anticipated future of autonomous vehicles has yet to materialize.

How do self-driving cars work?

Developing self-driving cars has been a goal for engineers for many years. The basic concept involves equipping a vehicle with cameras that can detect objects in the surrounding environment and programming in-car computers to follow the rules of the road and navigate to their destination.However, the reality is much more complex. Driving is a complex task that involves human judgment, reactions to weather conditions, and other decision-making skills that are challenging to program into a machine. Even tracking objects around the car on the road is more challenging than it seems.Companies like Waymo, a subsidiary of Google, use high-resolution cameras and lidar to estimate distances to objects by reflecting light and sound. The car's computers use this information to create a picture of the surrounding environment and anticipate how other objects might move. However, generating enough training data is difficult, so engineers must also rely on simulation data.In short, developing a self-driving car is far more complex than initially believed. Even the seemingly straightforward tasks hide a surprising amount of complexity. Therefore, it is understandable why self-driving cars are still not available for public use.

What a world with self-driving cars would be like? 

Despite setbacks, companies continue to invest in self-driving cars because they believe they will change the world and be profitable. Self-driving cars will allow passengers to read or sleep during their commute and could be used by ride-hailing services instead of human drivers. They could also greatly benefit people with disabilities who have difficulty getting around. However, experts disagree on whether self-driving cars will change car ownership in America. While some argue that people won't need to own a car, others believe that people will still want to own their own cars even with good ride-share coverage. Polls suggest that Americans are divided on whether they would use a self-driving car for their daily commute. The infrastructure will likely change over time to make it easier for self-driving cars to navigate, but this will require nationwide coordination and significant investment.

What are the reasons for the delay in getting self-driving cars on the road?

Self-driving cars rely on AI, and while the 2010s were a great decade for AI, with big advances in translation, speech generation, computer vision, and object recognition, AI still struggles to navigate our roads with the high degree of reliability required for self-driving cars.One of the major limitations is the need for vast amounts of training data. Training a self-driving car ideally requires billions of hours of footage of real driving, which is expensive and time-consuming. Additionally, since some events are rare, such as car accidents or debris on the road, the car may encounter situations that it has not been trained for.

Carmakers have attempted to address this by driving more miles, training cars in simulations, and engineering specific situations to generate more training data. While progress has been made, it has been slow, with Waymo currently testing cars on the streets of Arizona with no one behind the wheel, with plans to expand to more cities.

What are the safety concerns associated with self-driving cars? 

In March 2018, an Uber self-driving car, with a human safety driver behind the wheel, hit and killed a pedestrian in Arizona, USA. This incident highlights the fact that self-driving technology still has a long way to go. While some argue that fatal incidents are inevitable, as humans also cause fatal accidents while driving, it's important to note that human driving produces one fatal accident in every 100 million miles driven, whereas Waymo, the leading company in terms of miles driven, has only reached 20 million miles driven, making it too soon to determine if self-driving cars are as safe as or safer than human drivers. Uber has not driven nearly as many miles and has had a fatal incident. The company has not released specific figures, but it's uncertain if their driving record is worse than a human's without these numbers.Furthermore, a review of a pedisterian's death indicates that many preventable errors were made.

According to the National Transportation Safety Board, the near-range cameras and ultrasonic sensors were not in use at the time of the crash. The system also had issues with false alarms and was designed to assume that pedestrians would never cross without a crosswalk. In addition, the system failed to identify a pedestrian when he crossed without using a crosswalk and was unable to retain information about how she was moving, leading to a deadly collision.Despite Uber's response to the incident, implementing key safety improvements, sharing learnings with the self-driving industry, and accepting the NTSB's recommendation to implement a Safety Management System, deadly accidents with self-driving cars are likely to continue. Tesla's Autopilot system has also been implicated in a lethal 2018 accident, and three more recent deadly Tesla crashes have not yet undergone a full investigation. NTSB Chairman claims that drivers assume Autopilot lets them take their attention off the road when it shouldn't, presenting a potentially major problem.While self-driving cars have the potential to save many lives, significant engineering work is required to make them safe enough.

When will self-driving cars become a reality? 

While progress has been made, it's difficult to determine when fully autonomous vehicles will be available to the public. Companies such as Waymo and Cruise have been conducting test runs without drivers for several years, and Tesla's Autopilot system has improved with software updates, but full self-driving capabilities are still a ways off. Skeptics, such as Volkswagen's CEO, doubt that fully self-driving cars will ever come to fruition.Companies are hesitant to put their cars on the road until they're confident they're safe, as any accidents could be disastrous for their business. Although it's possible that self-driving cars could arrive in limited contexts this year, it's just as likely that timelines will be pushed back by several more years. So while self-driving cars are on the horizon, the question of when they'll actually arrive remains unanswered.