» Newsroom » Would you trust AI to drive a semi trailer?
Would you trust AI to drive a semi trailer?
Alexander Ollman enjoyed the Capstone engineering course so much that he took it twice. Now he's returned to Capstone as a mentor aiming to design a remote piloting system for the long haul trucking industry.
Wednesday 10 May 2023
A semitrailer rumbles over the Great Northern Highway. In the distance, a kangaroo bounds through the darkness, obscured by the glow of oncoming headlights.
A sudden course correction might be required. If so, who should make it? A human driver? Artificial intelligence (AI) powering an autonomous, self-driving system?
Or, a semi-autonomous piloting system controlled by a human operator thousands of kilometres away?
“Onboard AI performs billions of computations per second. Of course, it can spot road hazards and react faster than a human can,” said Alexander Ollman, founder of Totus Vehicle Control Systems, one of several companies developing technology-based solutions to address a global shortage of truck drivers.
“The problem is whether that reaction is the correct one. And, when I say correct, I mean: more human than not human.”
Ollman believes that people should be at the centre of every major technology innovation. His answer to the quandary above would be “both and” —a balance between humans and AI that Ollman and his Capstone engineering team are striving to perfect.
Feathers in our Cap(stone)
Ollman speaks with fervour about his undergraduate years in the School of Engineering at The Australian National University (ANU).
“There's no better feeling than graduating from university and going out into an industry that you are passionate about, and on your very first day realising: this is familiar. I understand what is going on here. In fact, I have a unique and creative way to solve the problems that this organisation has been trying to solve for a long time,” he said.
When Ollman arrived at ANU in 2015, he fancied himself a future “processor designer for Intel”, but he soon became captivated by the rewards and the possibilities of multi-disciplinary collaboration.
Along with the Systems Engineering framework that underpins all ANU engineering degrees, Ollman points to the Capstone engineering project as having inspired his philosophical transformation.
Capstone is a partnership between the School of Engineering and a host of industry clients including tech start-ups, engineering firms, and research and development teams. The industry professionals present potential projects to Capstone students on 'pitch day', after which the students self-select into teams.
As an undergraduate, Ollman enjoyed the Capstone course so much that he took it twice, acting as a proxy for the industry client the second time around. After a few years working as a Solutions Architect and Data & AI Specialist at Hewlett Packard Enterprise, he founded Totus and made enquiries about returning as a client.
He saw it as a way of giving back.
“The primary goal for me in introducing Totus as a Capstone project was to provide a beneficial learning experience and a fun learning experience,” he said. “I had no idea how they were going to react, how the university was going to react.”
Ollman prepared extensively for pitch day. There, he explained how the idea had come from a news report about a severe shortage of truck drivers in Australia, exacerbated by the pandemic.
As e-commerce and overnight delivery gained larger market shares, the logistics industry had been asking its workforce to push themselves to exhaustion, compromising road safety as well as the health of the drivers. As Ollman delved deeper into the issue, he learned the problem was not restricted to Australia.
“I saw a massive opportunity to help solve a problem and actively benefit so many people,” Ollman said. “It could allow truck drivers to work more flexible hours rather than days on the road away from family, and open the door for the new drivers the industry so desperately needs.”
A team of six ANU Capstone students agreed.
Start your engines
“The thing that jumped out at me on pitch day was the potential,” said Mitchell Grady, ANU engineering student with a Mechatronics major. “The final goal was stated clearly, and the client was happy for us to produce whatever we wanted that would contribute to that final goal.”
When 'Team TOTUS' decided they wanted to build a remote control steering wheel, accelerator and brake pedal, Ollman purchased a written-off 2014 Ford Focus which was garaged at one of the team member’s houses to await its transformation.
“One hard requirement I gave them was that a human driver could get in and out,” Ollman said. “They needed to design a solution where just one switch needs to be flicked or one piece of the equipment removed” so that the vehicle could be driven by a person inside the cab.
Borui Chen, a Renewable Energy Systems major, joined Team TOTUS in its second semester. Having been obsessed with remote control toys as a child, he found it thrilling to develop a full-size vehicle.
“I was excited about the project's goals and potential impact in the industry of autonomous driving, which is a heated topic nowadays,” Chen said.
Ollman met with the design team fortnightly. Because he had recently been a student himself, he remembered the courses that had built up to the Capstone project, and referenced them in his mentoring.
“He was super helpful with walking us through how to link all our Systems Engineering courses together and how to work through the full process, from concept to design to implementation,” said Grady.
Ollman was attentive and thorough when answering the students’ questions. But he made it a point to give them room to identify and try out solutions on their own.
“I put a lot of thought into how best to demonstrate systems frameworks and thinking, and how to map to roles, skills and projects they will perform in the industry when they graduate,” Ollman said.
For instance, he advised the team to seek out the input and expertise of the people who would be the end users and beneficiaries of the technology: truck drivers.
Like most ANU systems engineering graduates, Ollman believes a diversity in backgrounds and a diversity of thinking provides a competitive advantage in the engineering field.
“There are a number of problems in the world that, with the application of engineering thinking, could probably be solved very, very well. They can be solved even better if you also apply humanitarian thinking, artistic and creative thinking, and you draw from a variety of backgrounds and traditions,” Ollman said.
“That's when sparks fly, and great things are made.”
“While there may be differences in perspective and approach, what Capstone enables people to do is learn how to best leverage those differences and also how to best work with those differences, which is very analogous to real world environments.”
Problem solvers meet their match
Ollman considers himself a natural problem solver. Natural problem solvers, he said, tend to take degrees in engineering. And when Ollman researched previous attempts to design semi-autonomous trucks, he found that problems abounded.
“They could never make it seem safe 'enough' to satisfy investors, even if it was more than enough for the specific applications they were targeting,” Ollman said.
Also, if the trucks were to deliver goods over long distances, how would they be refuelled? If an engine overheated or a tire blew out, who would service it?
One answer that Team TOTUS came up with was a convoy of trucks with a mixture of in-person and remote drivers. There might be an in-person driver at the front and another at the back, for instance, while the other trucks would be piloted remotely.
Once the caravan had completed the long haul part of the journey, when more complicated suburban and city driving would be required, human drivers could jump behind the wheel and complete the routes.
“A particular operator doesn't have to leave their hometown,” Ollman said. “They pilot the truck remotely until it reaches a local headquarters, and then they or another driver take it from there.”
A fleet of trucks that can switch between in-person and remote operation would benefit the logistics industry, customers, and most of all drivers, who could spend more time at home with family and take better care of their health.
“The conversation about AI and machine learning should never be about replacing the human element,” Ollman said. “It should be about empowering and further enabling people with more options so they can do their job or whatever they're passionate about more effectively.”
Dodging the kangaroo
So, back to the kangaroo. How would an AI-assisted caravan react to a road hazard that appears very suddenly?
Ollman said the onboard AI systems will leverage a variety of sensors and other devices to constantly scan for hazards and calculate risk. Roos and other hazards will be detected well in advance, with a warning sent to all the drivers — remote and in-person — simultaneously.
“This information could be passed from the first truck to the rest of the convoy faster than a human driver could communicate it through voice transmission or brake lights,” Ollman said.
In certain situations, the AI system would make the convoy stop or course correct without consulting the human driver — for instance if there is no other way to avoid an accident.
“If I am a remote operator of the vehicle, I might say no, I want to be in control of the vehicle 100% of the time,” Ollman said.
Finding the optimal balance in human/AI relationships involves social, moral, ethical, legal, and philosophical discussions. All of this is within the scope of what Team TOTUS needed to research and did research.
“You can't just ignore these aspects of the system because they're not strictly an engineering problem,” he said. “And the amazing thing is, as systems engineers, we are taught those tools throughout our second and third year to investigate the possibilities and, based on the data available, chart a path to an optimal configuration, an optimal system.”
Proud dad moment
In the fall of 2022, Team TOTUS was among the many teams presenting their work at the ANU Capstone Showcase. The student teams designed schematics displayed on large poster boards, and those who had created prototypes brought them along for a “show and tell”.
Team SmartBin rolled in an industrial-size waste bin capable of measuring and reporting to a web interface the weight and density of its content. Not to be outdone, Team TOTUS unveiled a fully operational remote control driving system, while the adapted Ford Focus waited outside in case anyone wanted a test drive.
After the presentations, audience members — including academics, students, members of the industry, as well as a panel of judges — visited with each team and asked questions about their journeys and their findings.
The judges awarded three prizes, even though the plan had been for two.
When it was announced that Team TOTUS tied with Team SmartBin for the $1,000 first prize, Ollman applauded along with the rest of the audience as the students received their award certificates and posed for photographs.
“I think he wanted it to be our moment,” Grady explained. “We did end up getting a group photo with him later at the pub when we were all celebrating.”
In the photo, Ollman looks just as happy, if not more happy than the students whose work had earned the prize.
“Bit of a strange thing to say at my age, but it was very much a proud dad moment,” Ollman said. “To be announced as one of the winners meant an awful lot to the team. And as a result, it meant an awful lot to me.”
Chen now serves as Project Lead for Team TOTUS 2.0, a second team of Capstone engineering students who will further develop the project over two semesters.
“The win really boosted my confidence in continuing the project to provide more value for our client,” Chen said.