Self-Driving Cars – The difficulty of achieving the inevitable

Square

Self-driving cars: from 2020 you will become a permanent backseat driver

The Guardian 2015

In 2015, the Guardian predicted that by 2020, “you will become a permanent backseat driver” 2. Furthermore, Tesla founder Elon Musk had promised the world a completely autonomous car by the end of 2018 and when that was left unaccomplished it was pushed back to 2020 3. However, the question arises itself: what is being the blanket of darkness preventing this industry from reaching the glimmer of light behind? What problems and difficulties is the industry facing that are causing the accomplishment of this goal to be prolonged? Despite the extraordinary efforts by a myriad of leading tech companies and multitudes of resources, time and money being invested into this industry, the goal still stands unattained.

In order to understand and be able to answer these questions, we must travel back to the core underlying aspects of self-driving cars exploring what they are, their purpose and the techniques and developments that are being used in order for them to achieve that purpose. Only then can one really be able to identify the problems that come hand-in-hand with this goal.

A self-driving car, which is also known as an autonomous vehicle, is defined as a vehicle that is capable of being able to sense its environment and be able to navigate around that environment with no, or very little human input. The level of automation for a self-driving car is assessed by The Society of Automotive Engineers (SAE). In 2014, they published a classification system, J3016, Taxonomy and Definitions for Terms Related to On-Road Motor Vehicle Automated Driving Systems 4, consisting of 6 levels, from Level 0 to Level 5. Below is a visual chart that displays and explains each level, starting from no automation to full automation.

J3016, Taxonomy and Definitions for Terms Related to On-Road Motor Vehicle Automated Driving Systems 2

As of current, most autonomous systems are qualified as an SAE Level 2 (Partial Automation), meaning that the systems provide features for steering, brake or acceleration. However, a human must be present at all times to take control of the vehicle and these features must be constantly supervised. An example of this system is the Tesla Autopilot. There is, however, one system that is considered to be a Level 3 (Conditional Automation) and that is the Traffic Jam Pilot produced by Audi, but since its launch, the system has encountered manufacturing problems and delays obtaining regulation for production.

Self-driving cars employ on a wide range of technologies in order to carry out their functions. To start off with, these vehicles use a combination of sensors in order to be able to gain data and generate images about their immediate environment. These sensors include LiDAR (Light Detection and Ranging), ultrasound as well as cameras. Alongside this, they are fitted with a GPS (Global Positional System) unit in order to localise themselves and gain information about their current location. Furthermore, autonomous vehicles have been also known to include and use an inertial navigation system (INS) that makes use of accelerometers as well as gyroscopes to calculate the velocity, orientation and position of the vehicle.

The functions of autonomous cars as well as the technologies they employ on to achieve those functions 6

From this, the data, information and generated images are fed into the vehicles control system, which utilises image recognition along with neural networks and machine learning architecture in order analyse the data and hence navigate around the environment from this analysis.

Neural networks – Neural networks are a collection of algorithms that are modelled similarily to the human brain and are designed to be able to recognise relationships between sets of data and acquire the best possible result.

Machine Learning – Machine learning is a field of Artificial Intelligence(AI) that aims to provide systems the opportunity to learn and improve from data analysis.

Having now discussed the foundations behind autonomous vehicles, building upon this, we can start to identify what problems are being faced by the industry that is attempting to manufacture these vehicles that are fully autonomous. There are a potentially endless number of situations that the vehicle may encounter during its time of driving, and thus it must be prepared to be able to make safe decisions instantaneously with a high degree of accuracy in any given situation. Furthermore, these vehicles should also be able to conduct navigation in a various number of environments as well as conditions. With such high levels of variation that can occur in the environment, from different weather conditions such as torrential rain, ice and snow as well as an abundance of others to limitations in infrastructure, such as unpaved and unpainted roads, the difficulty of training the system architecture to be fully prepared for this is greatly increased and also the presence of certain weather conditions can result in reduced detection ranges of the sensors, making some obstacles potentially undetectable increasing the risks of incidents, regardless of how well trained the network is.

Continuing on, the industry has been interrupted by a bundle of incidents that have occurred during the operation of their autonomous systems by the general public. This has lead to increased criticism and questioning regarding the safety these vehicles provide. For example, on 18th March 2018, an autonomous car that was being operated by Uber hit and killed a pedestrian named Elaine Herzberg and it is believed to be the first fatality that involves a self-driving car. As a result, Uber then suspended its self-driving operations and engaged investigations into why this incident occurred. 6

It can be seen that, as of yet, the technology cannot be currently used safely and reliably to a point at which we trust it blindly. Although it is certain that the future of transport is heading towards driverless vehicles, this will not be an instant change but rather a timeline of developments that will be slowly inaugurated into our lives.


References

  1. [Featured Image] Ben Hernandez, “5G Technology Making Its Way Into Self-Driving Cars”, ETF Trends, Aug. 28, 2019, https://www.google.com/url?sa=i&url=https%3A%2F%2Fwww.etftrends.com%2Finnovative-etfs-channel%2F5g-technology-making-way-into-self-driving-cars%2F&psig=AOvVaw1O1ffWtsiU1OU5tH073Fnd&ust=1589218579906000&source=images&cd=vfe&ved=0CAIQjRxqFwoTCMDbz9zqqekCFQAAAAAdAAAAABAD
  2. Tim Adams, “Self-driving cars: from 2020 you will become a permanent backseat driver”, The Guardian, Sep. 13, 2015,
    https://www.theguardian.com/technology/2015/sep/13/self-driving-cars-bmw-google-2020-driving
  3. Aarian Marshall, “Elon Musk Promises a Really Truly Self-Driving Tesla in 2020”, WIRED, Feb. 19, 2019, https://www.wired.com/story/elon-musk-tesla-full-self-driving-2019-2020-promise/
  4. SAE International, “Taxonomy and Definitions for Terms Related to Driving Automation Systems for On-Road Motor Vehicles”, Jul. 06, 2018, https://www.sae.org/standards/content/j3016_201806/
  5. “Self-Driving Car Technology: How Do Self-Driving Cars Work? | Landmark Dividend”, Landmark Dividend, https://www.landmarkdividend.com/self-driving-car/
  6. Ron Schmelzer, “What Happens When Self-Driving Cars Kill People?”, Forbes, Sep. 26, 2019, https://www.forbes.com/sites/cognitiveworld/2019/09/26/what-happens-with-self-driving-cars-kill-people/#56b05adc405c

Author

Leave a Comment