Creating revolutionary new systems takes more than just engineering and marketing. Every aspect of the product has to be investigated first; that it is, what it is achieve, how it is to do it. Perfect cars don't exist, so we need to analyse what can go wrong, and how we can avoid problems. What about maintenance? Parts can wear out, get damaged, become obsolete so we have have to have plans in place for upgrading and replacing them. How will the product be used? Perhaps training modules must be created too. And what about the human element? People, also, are not perfect. Some make mistakes, a minority have evil intent. And, finally, how will motor insurance keep up?
Here at VK systems we have had more than two decades of collective experience in helping to design products as diverse as from washing machine seals to heat defectors for re-entry vehicles. Every project we take on is different and so requires a unique approach; an example of how we would look at a new challenge is detailed below.
Have you got a new, ground breaking idea? Do you need to talk to a highly experienced team of designers? Then get in touch with us today, initially by and we discuss your requirements informally, with no obligation on either side.
Tomorrow’s science often finds its origins in someone who is allowing their imagination to run free today. The automotive industry has come along way since the Model T Ford and with Satellite Navigation, cruise control, front and rear senses, to name but a few, technological advances, driving has become safer and more of a pleasure.
However, we do not yet have a totally safe, completely autonomous driverless car. How might we overcome the challenges that presents?
Perhaps the biggest challenge we face is in conceiving the task in hand. Why design cars that can drive like humans if they can be made so much better?
We are human beings, endowed with a brain so intricate that even neuroscientists don’t fully understand it. We apply our brains to thousands of different tasks each day, one of which may be driving.
In attempting to create a driverless car, our primary purpose to date has been to create a car that drives just like we do. This is perhaps our first mistake.
We might be good at driving, but we are imperfect, which is why we need car insurance, and we compensate for that by using our brains to get us out of difficulty, make snap judgement calls and use a range of strategies other drivers recognise, including hand gestures and eye contact. This enables us to avoid killing ourselves and other road users, although, every 100 million miles a fatality occurs.
The challenge is, how can artificial intelligence develop programmes that ape the very human characteristics of driving, the nods, the winks, the benefit of experience. Those complex social interactions that are so difficult for a robot.
How can we ensure that fully autonomous cars at worst allow no increase in the fatality rate and at best improves it markedly?
In solving this conundrum, we can look in two directions. Developing further the way in which semi-autonomous vehicles interact with each other and secondly, allowing AI to learn from not only the vehicle it is operating but all other vehicles which both store data in their on-board computers and send it back to a central repository.
The car must know where it is, where you want it to go and how to get there. Satellite navigation has made tremendous leaps since it was first invented although the readout is interpreted by us.
Even equipped with a compass, sextant and Lidar (light detection and ranging), enabling self-navigation on any road in the country, let alone the continent or world is a daunting prospect. It Doesn’t have to be, why would we need a self-drive car that can operate anywhere in the world?
The installation of complex three D maps that work with the vehicles sensors, enable the car to read the road and things in and around it, retaining and communicate back anything that has changed.
Zoning makes mapping achievable. Fully automated cars may mean that we dispense with the idea of personal ownership, because wherever we are a vehicle is available that navigates using a map of the area in which it operates. It doesn’t need a map of somewhere 200 miles away.
Traffic works to strict parameters, when these break down, are flouted by other drivers or are not appropriate, we act appropriately. In a traffic jam, we know that edging forward those few centimetres is essential to keep the traffic flowing, whereas AI intelligence might take the logical step to remain still until there is sufficient room to make moving forward worthwhile, something that could result in gridlock. But how do we instil that?
Perhaps the answer lies in thinking outside the box about future road design. If cars become truly autonomous and adept at communicating with each other, will we still traffic lights, pedestrian crossings, all kinds of markers and bollards etc. The key is perhaps to not only rethink the car but the environment it will operate in.
Research coming back from trials on semi-self-drive vehicles shows that they work best in good weather. Rain, fog, snow, frost and a host of other weather phenomena all have the capacity to affect performance. Our aim is a self-drive vehicle that operates in all conditions regardless with no scope for opting out. This is surely a matter of improving sensors, those that read the weather conditions and the ones that inform programme the car to make the necessary adjustments.
Improving radar and lidar are crucial tasks for us. The autonomous car will be highly dependent on radio wave signals which are emitted for several reasons, primarily to discover the distance of other vehicles and objects and calculate the time it will take to reach them.
There will be potentially hundreds of vehicles near a self-drive car, can we design systems that are fool proof and recognise only the responses to signals they have sent out? To make self-drive vehicles truly safe we need to be able to and the improvements we make need to incorporate anti-jamming mechanism or chaos will reign.
Performance and Day-to-Day Maintenance
We want our car to be able to read the environment and operate at a safe speed. In terms of performance, measuring and monitoring the use of the power available and the overall consumption of the fuel source, oil etc. the AI must manage the car so that it is always working as efficiently as it can be.
This isn’t too big a challenge, the cars we drive every day move ever closer to effective and economic system management and these can be further developed to provide the outcomes we want.
Self-diagnostics and self-adjustment where possible are also attributes we would want to see in a driverless car. The vehicle would be programmed to check its own systems and any aspects that can reduce efficient running or the safety of the vehicle. Where self-adjustment is not possible, an instruction would appear on the car’s computer screen.
Cybersecurity and Hackers
The car industry has only recently needed to start thinking about cybersecurity. We need to develop autonomous cars that are hack-proof so that passengers are safe and to prevent the vehicle from being used as a weapon. We must, in this endeavour take our lead from the major software companies who have already vast experience in cybersecurity.