On the Fast Track to Autonomous Vehicles: Accelerating The Future of Self-Driving Technology

 

For a driver stuck in a traffic jam, the sound of a distant ambulance or a police siren getting closer and closer is always a source of stress. But what if their car knew what to do and prompted them to steer calmly and quickly to a safe spot out of the way by using AI-powered smart city systems?

This happened in a recent test in Kaiserslautern, southwest Germany, as part of the AORTA project. The initiative brings together cities, universities, and industry partners to explore how the increasing number of autonomous vehicles will fit in with the trend toward highly connected smart city infrastructures. The project also examines how these vehicles will adapt to the evolving systems.

Autonomous vehicles use remote sensing technologies to monitor and map their surroundings, including road layouts, street markings, and people. Fast data processing enables these vehicles to make timely decisions about steering, speed, and other actions.

In the Kaiserslautern test, emergency vehicle AI-solutions were utilized wherein the vehicle sped through an intersection without stopping or slowing down. It took the shortest and quickest route thanks to the smart city infrastructure integration system developed by the AORTA team using artificial intelligence (AI).

The system uses micromanagement algorithms, perception technologies, V2X communication protocols, and traffic light control algorithms. It also contains precise information about the movements and locations of cars, pedestrians, and other road users at any given moment.

Vehicle-to-Everything (V2X) Technology Paves the Way for Autonomy

When the emergency vehicle needed to pass, autonomous cars stuck in the traffic lane could simply be steered out of the way by using V2X communication protocols. Cars driven by humans could be given quick, safe, and simple instructions to follow to achieve the same result. This saved precious seconds for the first responder.

“We’re quite proud of this solution, which is not just for autonomous cars but also cars driven by humans,” said Shawan Mohammed, R&D Program Manager, Automation, at Akkodis. “This is a worldwide first, automatically forming an emergency lane.”

Finding ways to safely and efficiently integrate increasing numbers of autonomous vehicles into our roads and cities is a key priority for software and technology experts. This aligns with the rise of smart cities and cognitive cities. Cognitive cities take the smart city concept a step further by enabling city infrastructure and institutions to use AI-driven transportation systems to communicate and interact with citizens.

Autonomous driving is about much more than driverless cars, explained Mohammed. “Often, people only see that we can drive these vehicles autonomously. If you reduce them to this fact, the obvious question is, why do we need such a technology?’

We need to make sure we use technologies to protect a cloud-based infrastructure that is digitalized from potential attackers.

Automation in Advancing Autonomous Vehicle Technology

Automation—and not just of cars—has an important role in society's shifting demographics. Many Western economies face an aging and dwindling population. Additionally, there is a growing need to carve out a technological niche as mass-market production moves to cheaper economies overseas.

AI and autonomous vehicle technologies are at least part of the answer, caring for unskilled roles such as delivery drivers. This allows humans to focus on more specialized, high-level roles that require the problem-solving abilities of the human brain.

Research and progress in the field of autonomous vehicles also help drive technology expertise overall, helping strategic sectors that demand overlapping skills, such as robotics.

Making the most of autonomous driving's potential will involve advances in areas such as the design of the vehicles themselves, their batteries, and the sensors and software that enable their operation. Also, it will be crucial to ensure their safe and secure operation in the face of cyber threats.

The VorSAFe-Plus project pushes the technological capabilities of autonomous vehicles to their very limits in the name of safety. It uses complicated algorithms to predict at which point an accident becomes unavoidable and decide whether to use autonomous vehicle safety systems such as automatic braking, steering, or deploying airbags to minimize the consequences.

The Akkodis team, in close collaboration with other project partners, is exploring how a dual-control system could take over from a human driver to prevent an accident. The system would override the limits set on ordinary cars driven by humans and carry out the kind of extreme maneuver that a human driver would not be able to pull off to keep occupants and those in the car’s surroundings safe.

Akkodis is helping to drive progress on all fronts, including one key obstacle to the broader use of autonomous vehicles: speeding up the development and testing process. The Akkodis team uses their technological and software expertise to move testing forward through a hybrid of real-world and virtual tests.

Shawan Mohammed, Akkodis R&D Program Manager, Automation

AI Testing Challenges

“The problem is that you need to drive around 4 billion kilometers to test the software stack or the AI of an autonomous vehicle,” Mohammed said. “You’d need hundreds of cars, driven for years, to guarantee their safety.”

With AI-driven autonomous vehicles, testing needs to be statistics-based to ensure every micro-change to a given scenario is thoroughly analyzed and tested.

As part of the IN2CCAM project, Akkodis' technology experts have been exploring virtual and hybrid testing solutions, which can accelerate that lengthy process while still ensuring safety. Testing projects taking place in several European locations, including most recently Vigo in Spain will lead to faster progress during the extensive testing needed during the cars’ development phase.

“In Vigo, in Spain, we have been testing autonomous driving in one lane of traffic, while the other lanes are used by normal traffic, for a kind of hybrid, or virtual, testing,” Mohammed said.

While the autonomous vehicle drives along the road, a pedestrian virtually “crosses” the street. “We cannot test this scenario with a real human since it's quite dangerous to do that, but we need to test it,” Mohammed said. “We need to understand how the car's behavior will be on real streets and how the AI inside the software stack will react in such a scenario.”

Autonomous vehicles are growing in number, and the smart city concept is developing in parallel. The two go hand-in-hand, Mohammed said. “Smart cities can help us accelerate this testing.” For the Akkodis autonomous vehicle experts, the focus in the following months and years will be more on testing and learning lessons from their existing projects that can help scale up the technology.

Mohammed shared that cybersecurity is a major focus area as autonomous vehicle expertise grows. “Security is one of the biggest showstoppers. We need to make sure we use technologies to protect a cloud-based infrastructure that is digitalized from potential attackers. As a follow-up project, we are thinking of looking into blockchain to have a centralized safety layer.”