150% rise in Midlanders who trust driverless cars

Public confidence in driverless cars is growing across the Midlands, with locals claiming they would feel “very safe” in a driverless car.

YouGov data analysed by Allen Leasing has revealed how Midlanders would feel about being a passenger in a driverless car on a main road in a big city.  The latest figures show that 5% of respondents would feel “very safe” in a driverless car compared to 2% in 2020.   

In contrast, some Midlanders are still sceptical as there was a 36.6% rise in those claiming they would feel “very unsafe” as a passenger in a driverless car with an increase from 30% to 41% in 2024. The percentage of Midlanders who would feel “fairly safe” decreased by –9.09%, declining from 20% to 19% over the past five years.  

Chanel Boddington of Allen Leasing commented, “As people take a step from science fiction into reality, it’s fascinating to see how attitudes toward driverless cars are evolving. While concerns about safety and reliability are natural, advancements in AI and sensor technology are helping to make autonomous vehicles safer and more efficient. As these cars become more common on our roads, greater awareness and real-world experience will play a key role in building public confidence." 

AI expert and scientific consultant to the Department of Transport, Professor Toby Breckon of Durham University shared his insights into the safety and technology of driverless cars. 

How Safe are Driverless Cars?  “Driverless cars have a range of very high-precision sensors onboard that give the vehicle a strong perception of the environment around it. 

Driving is about the perception of where you are and what is around you. We call it forecasting, which is what drivers do naturally as they predict what will happen next and so there is a lot of safety in that type of engineering solution. 

How do self-driving cars differentiate between objects like pedestrians, cyclists, and other vehicles in their environment? “All autonomous vehicles use cameras in form and many vehicles also use radar.

“There are radar units that we already have on most vehicles. Different vehicles use various sensors and onboard set ups, giving a 360-degree surround view of the environment simultaneously.

“Colour images are processed using AI algorithms to identify pedestrians, vehicles, cyclists, and different types of vehicles such as buses and lorries. The principle of autonomous driving is to drive where there is free and open space.

“The forecasting of what the objects are going to do in the environment based on their motion, characteristics, and knowledge of what they are, means that you can predict to a certain extent what is going to happen next. The key objective of self-driving cars is to transit into a free space where nothing is likely to transit to at the same point.  

What are the main differences between the AI used in autonomous vehicles and other types of AI, such as those used in recommendation systems or virtual assistants?  “Although they both rely on the core principles of AI called deep neural networks.

“The crucial difference is how it interprets all the sensor data onboard in real-time. 

A human driver's reaction time is typically 250 milliseconds, so a quarter of a second. An AI algorithm's reaction time will be faster than that. For example, your virtual assistant may take a few seconds to think about something and your vehicle hasn’t got seconds to do that.” 

What measures are in place to prevent hacking or malicious attacks on autonomous vehicles?  “Vehicles themselves are generally very secure and don't rely typically on information from the Internet.

“They rely on their perception of the world. The pathways from the sensor, especially to the processing on board are very secure. There are some potential issues around adversarial attacks, which involves creating signs in the environment that may confuse the vehicle.

“There's a current arms race between these techniques of those who invent them and those who mitigate against them. The reality is that when you're using multiple sensors, it's very difficult to perform an attack across all those modalities. 

“There's never been a major attack on the automatic update infrastructure for a major operating system, so for Apple or Microsoft or others, that infrastructure for the auto updates has never been successfully attacked. That's the type of infrastructure you would have for autonomous cars where they're downloading information.” 

What are the main differences between the AI used in autonomous vehicles and other types of AI, like those used in recommendation systems or virtual assistants?  

 “Across the AI systems, there are differences and there are differences internationally and regionally.

“Generally, I think it's perceived a lot of these systems will be capable of driving under a lot of different road conditions. There's no difference needed for, say, rural driving or urban driving.

“Typically, the AI algorithms are trained on data across a wide range of environments. As for different driving styles, I think that we'll probably see that drift away a bit.

“I think autonomous cars will have a particular driving style that moves, probably in caution and safety. Variations in driving style such as rapid acceleration will probably drift away because the system will be very safety conscious.”