European Guidelines Highlight Need to Regulate Truck AI

May 23, 2019 by Trucks.com, @trucksdotcom

By Sandeep Pandya

Editor’s note: Sandeep Pandya is president of Netradyne, a technology company that uses artificial intelligence, machine learning and edge computing to help reduce commercial vehicle crashes.

As it seeks to help businesses and organizations develop artificial intelligence that will be trustworthy and ethical, the European Commission has issued a document outlining rules and regulations for the technology’s development.
The guidelines, released in April, state that trustworthy AI should be:

  • Lawful, complying with all applicable laws and regulations
  • Ethical, ensuring adherence to ethical principles and values
  • Robust from both a technical and social perspective since, even with good intentions, AI systems can cause unintentional harm
Sandeep Pandya, president of Netradyne

Sandeep Pandya

This raises important questions about the global race to develop and deploy AI technology in virtually every machine. That includes, in our world, self-driving trucks. The commission’s mission is critical to ensuring harmony between people and machines. The technology must be developed in a way that serves humanity.

With the right intention and execution, regulations can push intertwined, global technological development forward. What do we need to consider for the trucking industry?

The ‘B’ Word – Bias

The success of this technology in the mainstream is highly dependent on people trusting and believing that AI is here to help. These regulations are a way of saying, “AI is here to stay, but how can we best harness it so we can flourish?”

This depends on the trust of human beings. Artificial Neural Networks (ANN) are a critical part of AI technology. This system of hardware and/or software is patterned to operate similarly to the human brain when it comes to decision making and processing data.

That said, a key part of this technology is its ability to adapt and modify as it learns from the world. For smart trucks, that means roadways and surrounding vehicles and traffic. One major issue with this is bias; how can we rid technology of bias if it is learning from a world filled with it?

For example, if a self-driving truck has to collide with more than one object, how does it decide which collision does the least harm?
In order to regulate bias within AI, you need more than just an ANN. You also need an additional layer of symbolic reasoning. For example, Netradyne’s Driveri can detect the flow of traffic and events, but also reason before passing judgement.

For instance, if a driver has a hard stop, could it have been because they were cut off by another vehicle? As we move forward and begin to integrate autonomous trucks onto the roadways, this type of higher level reasoning will be integral to removing bias from the technology.

Insurance and AI

About 2 percent of the total driving time for commercial drivers is considered hazardous. This 2 percent often defines a driver, especially when they are automatically assumed at fault in the event of a crash.

This is one reason more fleets are implementing platforms to monitor and protect drivers while utilizing AI to assign causality. That said, as we integrate AVs onto the roadways, the way we assign causality will have to keep up with new trends.

While the EU regulations don’t address insurance specifically, it will be important for future U.S. regulations to outline a roadmap for the way we assign fault for autonomous vehicles, especially as they share the road with their human counterparts.

This is just one reason why more fleets should adopt AI technology now. This will make easier transitions as these vehicles become more common.

4 stages of trucking autonomy

Four levels of truck autonomy. Guidelines for the use of AI could help ensure that safety and ethics are considered as technology advances. (Graphic: McKinsey & Co.)

 

Checks and Balances

Another critical issue raised by these regulations is that tension cannot exist between systems. Take autonomous trucks: An AI system turns on lights and another conserves energy. If a self-driving truck sets out on a long drive during the night, does the system in charge of turning on the lights overrule the energy conservation system?

Which system has the final say when both are designed for a specific function? How do we layer in the knowledge that these systems must work in unison to ensure a positive outcome?

And there is one question that these regulations don’t address that burdens the minds of many: Will AI be allowed to make mistakes and learn from them as humans do? Fleet managers can coach a human driver through mistakes to help ensure they don’t happen in the future.
AI technology can learn as well. But will we allow it to do so or, rather, expect the technology to be perfect from the get go?

Will the U.S. Be Next?

The European regulations raise questions about what this means for the transportation industry in the U.S. Will everyone follow suit, or will these guidelines be taken as that – guidelines – not law? The U.S. has released a national strategy surrounding the growth and development of this technology, but as a free market society, isn’t leading the charge when it comes to regulation.

This is a complex issue. Regulations are critical for general safety but often restrict innovation. In the past, we have seen major players within the space band together to regulate themselves. The aviation industry competes fiercely in most areas. But there’s an established standard of collaboration and transparency when it comes to risk and safety developments.

The transportation industry may want to adopt this pattern with AI. Companies leading the charge will have to self-regulate together for the good of others. The Society of Automotive Engineers could take the lead in regulating AI within the trucking and autonomous vehicle space.

Could Regulations Hurt the Trucking Industry?

Although these regulations were thoughtfully drafted to ensure the European Union develops and moves this technology in the right direction, they could have adverse effects. When regulations aren’t the global standard, those with regulations end up competing with those without. That often gives the advantage to the unregulated.

Transparency, a huge factor in this area, is a common theme throughout the European Commission’s guidelines. Organizations that do not wish to give up their “secret sauce” may just opt out of deals with those whose nations require it. That dictates the need for input from both government and industry leaders to create successful and effective guidelines.

Editor’s note: Sandeep Pandya is president of Netradyne, a technology company that uses artificial intelligence, machine learning and edge computing to help reduce commercial vehicle crashes.

Trucks.com welcomes divergent thoughts and opinions on transport technology and trucking industry issues. Use the comments section to cite yours. Qualified opinion leaders are welcome to offer suggestions for opinion columns. Contact info@trucks.com.

Trucks.com April 16, 2019
A growth in autonomy brings benefits, but also the risk of cybersecurity breaches, which grow as human intervention with a truck’s controls lessens.

Leave a Comment

Your email address will not be published.

×

Subscribe to our mailing lists

Choose one or more topics:
x