Self Driving Car Safety: Is Driving Safer?
Self driving car safety is at the forefront of consumer concerns and ethical debates when it comes to driverless cars. Leading the way are companies such as Google, Uber, Audi, Delphi Automotive, Tesla, Nissan, and Mercedes-Benz in self driving car technology. Driverless safety technology will feature automated technology using an array of sensors and computerized cameras, lidar and radar to reduce accidents by up to 90%. The software will process data used by these sensors to provide a safe and efficient journey.
A Glimpse at Self Driving Car Safety Features
Autonomous vehicles, such as Google’s self driving car, are equipped with advanced GPS technology that allows them to know their exact location and driving environment due to detailed mapping. Other technologies in Google’s self driving car include numerous sensors, lidar, radar, cameras, and adaptive software to sense obstacles and other vehicles.
Together, these sensing technologies input data into the software, which operates the vehicle at a safe speed, with the ability to accelerate, brake and make turns. The vehicle is able to do this while sensing obstacles such as pedestrians and cyclists, as well as other vehicles. The vehicle combines all of this and plans its movements and routes accordingly.
Advanced Driver Assistance Systems are examples of self driving car safety features being utilized by major automotive manufacturers today, and the 2015 Chicago Auto Show is a prime example in showcasing the latest in driverless car safety. While these technologies still allow human intervention, it does lift the cutain for what drivers can expect in the future.
Advanced Driver Assistance Systems feature an array of various technologies, depending on the vehicle. A feature such as forward collision avoidance makes use of forward facing sensors, consisting of radar, laser or cameras. These sensors work to apply or increase the brake to help prevent human error. Brake assist works to improve braking response time by pre filling the brake pumps. Similar to these features is automatic warning and braking, which uses cameras, lasers and radar to alert the driver to obstacles or the need to brake. Depending on the vehicle, autonomous braking may activate if driver response time is slow.
Backup cameras are growing to be a necessary safety feature, with the NHTSA issuing a statement that every vehicle sold after May 1, 2018 will need to be equipped with one. These cameras are attached to sensors that alert drivers if they are at risk for hitting an object.
Vehicle-to-Vehicle Communication will allow for self driving cars to sense the speed and path of other vehicles. Features such as adaptive headlights are proof that vehicle-to-vehicle communication is possible, by adapting headlight brightness for oncoming traffic and when following vehicles. Lane Detection is a feature that utilizes cameras and sensors to monitor the vehicle in the lane. These camera and laser sensor systems work together to help drivers from drifting out of their lane. Certain Mercedes-Benz and Volkswagen models autonomously steer, change lanes and swerve if needed to avoid accidents. Volvo offers Pedestrian and Cyclist Detection which is an automatic braking system to avoid obstacles that come into the vehicle’s lane. Blind spot alert works similar to this, with sensors monitoring blind spots and rear cross-traffic.
Driverless car safety technology also goes as far as assisting in parking, with self-parking features. Fiat Chrysler Automotive introduced its ParkSense technology, which is a parking assist system that helps drivers with parallel and perpendicular parking. Using the rear fascia’s ultrasonic sensors, the technology can detect objects and obstacles from 10-59 inches away. The vehicle will display both a visual and audio warning. As this technology progresses to fully autonomous parking, it will make the act of parking and parking space itself more efficient. Fully self driving vehicles will be able to drop passengers off at the door, and park closer to other cars as there is no need for open door space. McKinsey & Company reports that this could save 5.7 billion square meters in parking space, as parking will be 15% tighter.
Mobileeye is an example of being part of the evolution of self driving car safety technology, providing machine vision technology with mono-cameras. The system’s mono-camera inputs data to system’s OEM, which vigorously processes the data. The processing process is comprised of several features: processing modules, general purpose CPUs, and IO capability. Features such as Pedestrian Collision Warning, Lane Departure, Forward Collision Warning, Speed Limit Indication, and more come from the data driven through Mobileeye’s system.
The popularization of self driving car safety features prompted the National Highway Traffic Safety Administration (NHTSA) to create a rating system which ranks vehicles based on their autonomous safety features. Level 0 means that human control is needed for the entire ride, whereas Level 4 is equipped with in-vehicle-crash avoidance systems that perform all preventative functions for the entire duration of the trip, without any human intervention. A level 1 classification means that controls are operated independently of one another. For a vehicle to classify as a level 2, a minimum of two computerized controls have to work in unison. An example of this could be seen in adaptive cruise control with lane keeping warning. Level 3 classification requires drivers to allow self driving car safety technology to take control, sensing its surroundings and the conditions at hand then alert the driver when it’s time to take over.
What Are The Risks With Self Driving Cars?
Many experts agree, there is no such thing as zero risk. Google’s self driving car has only been in 13 minor accidents in its near 2 million miles of driving tests. In these 13 incidents, the self driving car was never at fault. While the risk of driver error may be removed with autonomous technology and lack of need for human intervention, self driving cars will still be susceptible to risk through its computerized systems, networks, and sensors.
As a result, self driving car safety technology faces certain hurdles that automakers will have to conquer. For instance, environmental conditions such as heavy snow may affect the sensors. To combat these potential problems, driver intervention will be necessary. Drivers will not be able to completely rely on these sensors and still need to be alert of these issues. As vehicles progress to being fully autonomous, vehicular technology will have to compensate and find a way around these issues.
As self driving cars become completely driverless, they will be forced to rely on computers which could be susceptible to data threats and failures. One failure could be lacking the ability to read qualitative gestures. For instance, if a traffic cop is directing traffic, the self driving car would not be able to read their hand signals. Current sensors and technology are not equipped to recognize these movements. If vehicles became completely autonomous, self driving car safety functions would need to be built around the qualitative aspects of driving. It could also require the traffic cop to utilize a system of automated controls that could communicate commands to the vehicle and direct its behaviour.
Car Hacking: A Digital Threat
To optimize self driving car safety, safety features utilize the vehicle’s networks to alert drivers and implement their respective functions. Wireless networks are used to connect the vehicle’s Electronic Control Unit. The Electronic Control Unit regulates the vehicle’s braking system, locking mechanisms, the engine itself, and more. This system allows for mobile devices to sync and control various features. However, relying on computerized systems can be risky, as computer failure could be hazardous or even fatal. As well, some experts indicate that the real self driving car safety issues lie within the gathered data and the privacy of that data.
Digital security precautions are necessary to avoid computer tampering, hacking and failure. Vehicles equipped with a USB port could be tampered with by simply plugging in a USB mouse or keyboard, in which you can interact with the vehicle’s screen. Although the likelihood of a digital attack is improbable, if the vehicle is not equipped with the proper security software it could pose a potential risk.
Some vehicles, such as the Tesla Model S, feature a hidden ethernet port. The port operates the vehicle’s network, which consists of the dash and navigation screens, centre console and other functions. This could allow for a hacker to remove software restrictions and tamper with the current software. Whether it is customizing the software to better suit the driver’s needs or maliciously influencing its function, it could pose a risk to the vehicle’s overall operation.
It is also possible to wirelessly hack a vehicle. Most vehicles today come equipped with Bluetooth connectivity. Car hackers have the potential to utilize this technology by remotely accessing the driver’s smartphone and install an app with a virus, as tested and proven by a team of researchers from the University of Washington and University of California. The team of researchers were able to prove that by simply coding an audio CD with malicious content and playing it in the vehicle’s sound system could pose a threat. Researchers have also found that hacking into a vehicle could allow the hackers to adjust the fuel gauge, speedometer, as well as honk the vehicle’s horn, pull the steering wheel back and forth, and immobilize the brakes. Results can be worsened if the hacker(s) travel beside the vehicle they are trying to attack.
Hacking threats have been proven, when hackers hacked a Jeep Grand Cherokee and were able to remotely disable and manipulate its functions. Andy Greenberg served as the driver in a hacking demonstration, put on by Charlie Miller and Chris Valasek. Greenberg was not aware of what the hackers were going to do to the vehicle prior to their attack. It began with the manipulation of temperature controls, then the radio station changed and the volume increased. The control knob to turn the music off was disabled. Next, windshield wipers and wiper fluid were initiated. Appearing on the infotainment centre’s screen was a photo of the hackers. The hackers were also able to manipulate the vehicle’s ability to accelerate, enable and disable the vehicle’s ignition, knock out the braking system, as well as track the vehicle’s GPS coordinates. The hackers could also pinpoint the vehicle’s IP address for continued access from all over in the country. Miller and Valasek believe that by releasing the information on how they were able to hack the Jeep will help automotive manufacturers better secure their networks. The scariest part of car hacking is that the results could be fatal.
As a result, Fiat Chrysler Automotive had to issue a safety recall to update the software, preventing further wireless attacks. Issues such as this bring on real cause for concern. Security expert Josh Corman cautions that automotive manufacturers may not be able to protect susceptible features. He states that “if it takes a year to introduce a new hackable feature, then it takes [automotive manufacturers] four to five years to protect it.”
Some experts believe in order to prevent car hacking and increase self driving car safety, federal regulations will need to be implemented to protect vehicle owners, which demand advancements in automotive digital security as safety and hacking safeguards will be necessary.
Are Self Driving Cars Actually Safer?
Risks in terms of self driving car safety concern many, due to the fact that there will be limited or no ability for human intervention if something were to go wrong. However, self driving cars have great potential to reduce accidents caused by human error.
Apart from aiding in the reduction of accidents, self driving cars will also have the potential to reduce traffic and congestion on roadways. This is because driverless cars will have the capacity to not only drive closer to other cars thanks to their advanced sensors, they will also be able to travel at a higher speed due to increased reaction times and reduction in erratic behaviour of driven cars as the proportion of autonomous vehicles increases.
McKinsey & Company reports that “for every person killed in a motor-vehicle accident, 8 are hospitalized, and 100 are treated and released from emergency rooms.” On top of that the Eno Center for Transportation reports that for every 10 accidents, 9 are due to human error. Experts such as Ryan Hagemann, fellow on robotics at think tank TechFreedom, believe that self driving car safety technology could help reduce accidents by up to 99.9%, with the right precautions and risk management protocols. If safer is better, self driving car safety is leading the way.