Semantron 22 Summer 2022

Autonomous vehicles

of individuals, would not have the resources to sustain their appeal. It is therefore necessary for courts to design separate legislations specifically for autonomous accidents, where the criteria for certain types of cases are written in stone with no room for interpretation. This will of course take years to develop and pass through governmental bodies.

Conclusion

This essay has reviewed the various hardware and software that an autonomous vehicle uses, whilst applying these technologies to the decision-making process that occurs in a self- driving car’s onb oard computer using algorithms and DNNs. The essay has also reviewed the prospects of a fully autonomous network and has presented both sides of the argument when answering the question: ‘How likely are we to achieve a fully autonomous network in the next decade?’ However, as discussed, an extensive range of technology is used for an autonomous car to function. A self-driving car uses RADAR, LIDAR, Ultrasonic sensors, GPS, cameras, and infrared sensors just to see where it is going, let alone understanding the objects around it. Even with this multitude of hardware, variable weather conditions such as fog or even the sun’s glare can greatly limit the performance of the sensors. Additionally, the algorithms and DNNs needed to process the information provided by the hardware and to identify the surroundings of the vehicle are far from robust; the algorithms are not reliable and developed enough to identify objects quickly and correctly, while still needing to calculate the optimal solution for how the vehicle to react to the situation. Moreover, this whole process must take less than 1 second, not only to surpass the brake reaction time of a human, but more importantly, to avoid a potentially fatal crash. The development and improvement needed in algorithms will take an immense amount of time to achieve to ensure that the reliability and uncertainty in the decision-making process is at an acceptable level. But the technical aspects of a self-driving car are not the only reason why a fully autonomous network will not be achievable in the next decade. The survey I conducted has illustrated how the public will most likely be averse to the idea that they are no longer in control of where the car is going, while being sceptical of the computer’s ability to make the ri ght decisions. As with almost everything, autonomy will never succeed if it does not have the trust and desire of the public, and it will not take a year or two in order to convince the public that autonomy is better than driving: manufacturers will have to demonstrate consistent results to prove to the government that their vehicles are safe, let alone the citizens who simply might reject the idea of autonomous travel all together. The ethics and moral compass of the self-driving vehicles will be fiercely disputed amongst heads of departments within the manufacturers, members of parliament, and the public. This transition and reform of legislation regarding autonomy will be a long and tedious process as several authoritative bodies are required to cooperate and find a solution which adheres to public safety. Furthermore, the logistics of a fully integrated autonomous network will take years to implement, but more importantly, money. Governments may not have the money to build or develop new infrastructure which accommodates V2I and V2V, and even if they did, would it be in the government ’ s best interest to pour millions, if not, billions of pounds into an autonomous network, when that money could be spent elsewhere? Because of the pandemic, companies will be reluctant to invest in projects or initiatives as big, as expensive, and as risky as an autonomous network. Brexit will also play a major role in the development of autonomy in the UK. All the major car manufacturers who specialize in autonomy/electric vehicles ,such as Toyota, Waymo, and Tesla, come from across the world. This will

31

Made with FlippingBook interactive PDF creator