Tesla has grand plans for the Autopilot computer system in its cars, but how realistic are the firm’s self-driving vehicle ambitions?


SEE: Elon Musk and the cult of Tesla: How a tech startup rattled the auto industry to its core (free PDF) (TechRepublic cover story)

What is Tesla’s Autopilot?

Autopilot is an optional driver-assist system for Tesla cars that Tesla CEO Elon Musk has promised will eventually turn the electric cars into fully self-driving vehicles.

When Autopilot is engaged, cars can self-steer, adjust speed, detect nearby obstacles, apply brakes, and park. The technology uses a combination of radar, cameras, ultrasonic sensors, and GPS.

Tesla says the onboard computers in its Autopilot-enabled cars released since October 2016 can support full self-driving capabilities, and that this functionality will be added via firmware updates over time, subject to regulatory approval.

Note: TechRepublic’s first Tesla Autopilot cheat sheet was originally published in July 2016.

SEE: Autonomous vehicles and the enterprise (ZDNet/TechRepublic special feature) | Download the free PDF version (TechRepublic)

Does Autopilot turn Teslas into fully self-driving cars?

Not yet. Currently, Autopilot is not rated as a fully self-driving system.

Instead, Autopilot is currently classed as a Level 2 automated system by the US National Highway Transportation Safety Administration. Level 2 designation means Autopilot is capable of “combined automated functions, like acceleration and steering” but that the driver must remain engaged with driving at all times.

SEE: All of TechRepublic’s cheat sheets and smart person’s guides

What can Tesla’s Autopilot do?

At present, the latest version of the Autopilot software, known as Enhanced Autopilot, offers driver-assist features mainly designed to help with highway driving. These features include:

  • Traffic-Aware Cruise Control: Drives at the current speed limit, keeping a safe distance from the car ahead and adjusting speed for bends or cars passing in front. It is able to observe two cars in front of the vehicle.
  • Autosteer: This feature keeps the car in the current lane and engages Traffic-Aware Cruise Control to maintain the car’s speed, up to 90mph. It relies on various measures, including steering angle, steering rate, and speed, to gauge how to control the car. Drivers must keep their hands on the steering wheel, or else an alert goes off.
  • Auto Lane Change: Enables the Tesla to automatically change lanes when Autosteer is engaged. Switching on the turning signal will cause the Tesla to move to the adjacent lane when safe.
  • Side Collision Warning: This alerts drivers to objects, such as cars, that are too close to the side of the Tesla. When the car detects an object close to its side, fluid lines will radiate from the car’s image in the Instrument Panel to alert the driver.
  • Autopark: Automatic parallel parking, including the ability to automatically spot parking spots when driving at low speeds around cities.
  • Summon: Allows a phone app to automatically park a Tesla in or retrieve it from a home garage or driveway, and can also be set up to automatically control garage doors.

Other features include Automatic Emergency Braking and Steering, Automatic High Beams, Blind Spot Detection, and Speed Assist.

Those who don’t want to pay for Autopilot, who chose not to pay for the Enhanced Autopilot package, can only use safety features such as lane-departure warning, automatic emergency braking, and collision warning.

SEE: Tech and the future of transportation (ZDNet special report) | Download the report as a PDF (TechRepublic)

Which sensors does Tesla’s Autopilot use?

Tesla says Autopilot’s sensors allow the onboard computer system to build a detailed picture of its surroundings, allowing the vehicle to anticipate possible collisions with vehicles, pedestrians, cyclists, animals, debris, and other obstacles. It can also detect road markings, signs, and traffic lights.

For drivers that opt for the Enhanced Autopilot and Full Self-Driving Capability, systems in cars launched since October 2016 rely on the following sensors.

Forward-looking radar: The radar used by Autopilot can see up to 160m ahead of the car, through “sand, snow, fog–almost anything,” according to Musk. Radar is the primary sensor used to detect the vehicle’s surroundings, along with the front-facing cameras.

Eight cameras: The four forward-facing cameras on the windshield of the car serve as a backup to the radar. The cameras consist of a narrow camera that captures footage 250m in front, a main camera that captures 150m in front, a wide-angle camera that captures 60m in front, and a camera that captures footage 80m in front and to the side of the car. The wide-angle camera is designed to read road signs and traffic lights, allowing the car to react accordingly, however, there is debate over whether this feature is enabled in cars with Autopilot 2.0 hardware. A pair of rear cameras captures footage up to 100 meters to the rear and the rear sides of the car.

Sonar: A 360-degree, ultrasonic sonar detects obstacles in an eight-meter radius around the car. The ultrasonic sensors can spot objects like a child or a dog, and work at any speed. This feature can also detect objects in blind spots and assist the car when automatically switching lanes.

GPS: A satellite navigation system can detect the car’s position on the road.

Image: Tesla

Additional resources:

Which computer hardware does Tesla’s Autopilot run on?

Today, all Tesla vehicles–Model S, Model X, and the new Model 3–are equipped with a custom-version of the Nvidia Drive PX2 platform.

The computer delivers more than 40 times the processing power of the system used in earlier cars and runs a Tesla-developed neural net that “sees” the world around it by interpreting data collected by the cameras, radar, and ultrasonic sensors.

There have been major updates to the onboard computer and sensor array used to run Autopilot since its launch, with revisions including AutoPilot 1.0, Autopilot 2.0, and Autopilot 2.5.

Tesla says since Autopilot 2.0 was introduced in October 2016 all cars that opt into Enhanced Autopilot and the Full Self-Driving Capability “have the hardware needed for full self-driving capability at a safety level substantially greater than that of a human driver”.

However, since making this claim, Tesla has still upgraded Autopilot’s underlying computer hardware, and in mid-2017 it began fitting cars with Autopilot 2.5 hardware, which adds an additional system-on-a-chip (SoC) to the custom Nvidia Drive PX2 platform used in Tesla cars.

Tesla hasn’t confirmed which chip was added, although it is rumored to be an Nvidia Parker SoC, but has described the addition as a minor change. Tesla is also working on its own custom AI chip that Musk says will be able to analyze 10x more frames of video per second than the existing Nvidia platform, and which will feature in upcoming Autopilot 3.0 hardware. Despite these planned upgrades for future cars, Tesla has previously said that, in the “highly unlikely” event that Autopilot 2.0 hardware can’t support fully autonomous driving, it will upgrade cars at no additional cost if owners had opted for the full self-driving package.

There is also some debate among Tesla owners over whether Autopilot has lost, as well as gained, capabilities over time.

A Tesla employee drives a Model S electric automobile, equipped with an early Autopilot version on a highway in Amsterdam, the Netherlands.
Image: Jasper Juinen, Bloomberg via Getty Images

Will Tesla cars ever become fully self-driving?

Some automotive industry experts have questioned whether a Tesla would be capable of becoming fully autonomous driving without using a lidar, a sensor which measures distances between objects using pulses of light.

Many other self-driving cars, such as those made by Waymo, rely on lidar to provide precise distance measurements and high-resolution maps of surrounding objects. Musk has argued lidar serves only as a “crutch”, and that it’s possible for driverless cars to rely solely on radar, cameras, ultrasonic sensors, and GPS, as is the case in Tesla vehicles.

Additional resources:

Which cars is Tesla’s Autopilot available for?

Autopilot is available for the Tesla models S, X, and 3, and will likely be used in all future cars released by Tesla.

How many crashes have there been involving Tesla’s Autopilot?

Since launch, there have been several crashes, including fatalities, involving Teslas using Autopilot.

The first fatal crash involving a Tesla using Autopilot occurred in May 2016, when a Model S collided with the trailer of a truck turning left across the path of the car.

At the time Tesla issued a statement indicating the accident resulted from a series of highly unlikely circumstances.

“Neither Autopilot nor the driver noticed the white side of the tractor trailer against a brightly lit sky, so the brake was not applied,” it stated.

“The high ride height of the trailer combined with its positioning across the road and the extremely rare circumstances of the impact caused the Model S to pass under the trailer, with the bottom of the trailer impacting the windshield of the Model S. Had the Model S impacted the front or rear of the trailer, even at high speed, its advanced crash safety system would likely have prevented serious injury as it has in numerous other similar incidents.”

In its report into the collision, the U.S. National Highway Traffic Safety Administration (NHTSA) Office of Defects Investigations concluded that Autopilot and other onboard safety systems were working at the time of the crash, going on to say: “Although these technologies had limitations, the ADAS [Advanced Driver Assistance System] system did not respond to an impending crash event.

“Regardless of the operational status of the Tesla’s ADAS technologies, the driver was still responsible for maintaining ultimate control of the vehicle. All evidence and data gathered concluded that the driver neglected to maintain complete control of the Tesla leading up to the crash.”

More recently, a Tesla Model X using Autopilot was found by the US National Transportation Safety Board (NTSB) to have been steered into a crash barrier while traveling at just over 70mph on a US Highway 101 in Mountain View, California.

The NTSB reported the vehicle didn’t detect the driver’s hands on the wheel for six seconds prior to the crash. It found the car was steered to the left seven seconds before impact, and, rather than braking or taking evasive maneuvers, the vehicle accelerated in those final seconds.

In the wake of the crash, Tesla said it was “working closely with investigators to understand what happened, and what we can do to prevent this from happening in the future”.

It also released statistics that there is one automotive fatality every 86 million miles across all vehicles from all manufacturers, going on to state “for Tesla, there is one fatality, including known pedestrian fatalities, every 320 million miles in vehicles equipped with Autopilot hardware. If you are driving a Tesla equipped with Autopilot hardware, you are 3.7 times less likely to be involved in a fatal accident”.

Additional resources:

When will Autopilot turn Teslas into self-driving cars?

There’s a big question mark over when, given there are no fully self-driving cars certified to operate without any human assistance today.

Elon Musk recently said that Tesla’s plan to demonstrate Autopilot’s full self-driving abilities had been put on hold to allow the Autopilot team to focus on safety and existing features.

SEE: The new commute: How driverless cars, hyperloop, and drones will change our travel plans (free PDF) (TechRepublic cover story)

Although the technology that enables self-driving cars–the radar, cameras, GPS, lidar, and machine-learning systems–have become much more capable and affordable in recent years, many automakers are more conservative in estimating how quickly self-driving cars will be developed, with most planning to have a semi-autonomous vehicle available no earlier than 2020.

Additional resources:

What other barriers are there to a self-driving Tesla?

Even if Tesla were to develop the technology to support fully self-driving cars there would still be regulatory roadblocks. Some US states currently have specific laws that would ban autonomous driving. Without clear regulations, testing self-driving cars is a challenge.

Alabama, Arkansas, California, Colorado, Connecticut, Florida, Georgia, Illinois, Indiana, Kentucky, Louisiana, Maine, Michigan, Mississippi, Nebraska, New York, Nevada, North Carolina, North Dakota, Oregon, Pennsylvania, South Carolina, Tennessee, Texas, Utah, Virginia, Vermont, Washington and Wisconsin–and Washington D.C.–have all passed legislation related to autonomous vehicles, as of July 2018.

According to Jeffrey Miller, IEEE member and associate professor of engineering at the University of Southern California, that progress is partly a result of lobbying pressure from Silicon Valley and auto companies in the state.

Who are Autopilot’s competitors?

Tesla is not the only car manufacturer hoping to build self-driving cars. Several large automakers, including Ford, Volvo, and Toyota, have announced plans to develop AI systems capable of autonomous driving, and companies such as Waymo, and to a lesser extent Uber, have extensive autonomous car programs.

How much does Tesla’s Autopilot cost?

In addition to the cost of the car, Tesla charges $6,000 for Autopilot’s driver assist features–known as Enhanced Autopilot–and $5,000 for the “full self-driving capability” Tesla says will eventually be enabled.

Additional resources:

Subscribe to the Innovation Insider Newsletter

Catch up on the latest tech innovations that are changing the world, including IoT, 5G, the latest about phones, security, smart cities, AI, robotics, and more. Delivered Tuesdays and Fridays

Subscribe to the Innovation Insider Newsletter

Catch up on the latest tech innovations that are changing the world, including IoT, 5G, the latest about phones, security, smart cities, AI, robotics, and more. Delivered Tuesdays and Fridays