Zuduo Zheng: CAV man
Associate Professor Zuduo Zheng hails from the School of Civil Engineering, at the University of Queensland, and a DECRA Research Fellow sponsored by the Australian Research Council. His research zeroes in on transport engineering – traffic flow modelling, travel behaviour and decision making, advanced data analysis techniques (mathematical modelling, econometrics, and numerical optimisation) as well as meta-research.
What are you working on at the moment?
I’m working with a group of highly talented researchers to try to develop a comprehensive model that should be capable of simulating mixed traffic on our roads – that’s conventional, connected and automated vehicles – to the microscopic level. We have an Australian Research Council grant to fund this.
We’ve spent more than two years first designing an experiment using an advanced simulator. We needed to get data about connected vehicles, but there’s insufficient data out there for researchers to use. We developed a new model for conventional, connected, and automated vehicles. The results are positive, but we’re only 80% done. The next phase is to secure financial support to test the model with real data – that is on a test site.
What was the ‘ah ha’ moment when you decided this would be your career?
I didn’t decide on this field until I was doing my Master’s Degree research. But I do remember in my second year of high school, I was trying all kinds of crazy animated ideas and my chemistry teacher told me maybe I could become a good researcher. I was kind of upset by that comment, I thought, ‘You think I’m boring.’ But, now I look back and really appreciate this comment.
What does the average person not know about human behaviours and decision making when it comes to self-driving cars?
There are misconceptions about connected vehicles and what they do. They have some kind of communication capability so they can either talk to other vehicles to exchange information in regard to congestion or roadwork zones ahead, also some vehicles can communicate with infrastructure.
There’s a big misconception when it comes to the difference between connected and automated vehicles. People often talk about automated vehicles as if they are connected vehicles – they can be totally different types. For most of the automated vehicles being tested, they have limited or no connectivity at all, so they can’t talk to other vehicles. In an ideal situation, automated cars should have this connectivity.
Public opinion about automated vehicles is generally positive, based on recent surveys conducted in Australia and overseas, but when it comes to driver behaviour and decision making in those vehicles, I’m not sure what would be important factors the average person really cares about … yet. Many factors have been reported, such as energy efficiency, environmental protection, safety performance, etc. But factors they stated they care about can be very different from factors they consider when they are actually buying a new vehicle. It may simply come down to the price tag.
Hypothetical time. Someone’s given you a huge bucket of money and a reasonable time to do it. Think of a problem you’d like to fix, and why.
I’d want to look at the issues related to mixed traffic – conventional, connected and automated vehicles – from a system level, that is putting them all together on the road. That means investigating the communication, connections, mathematical modelling, human factors, and all the different issues using a holistic approach.
These days, mainly because of budget constraints, researchers are forced to look at these issues separately, but we often tend to underestimate the complexity and overestimate the benefits when we do so. For instance, five years ago, many people optimistically predicted driverless cars being in the market by 2020 and being the mainstream by 2030. But we’ve gradually realised there are more challenges to sort out first.
To me, this is a consequence of not looking at all of the components together.
Hypothetical number two. This time with a small budget and timeframe. What would you choose to have a go at in this scenario?
Very limited budgets are happening all the time. I’d like to focus on developing microscopic models suitable to reproduce some important features when we mix together connected, conventional and automated vehicles. These are critical for the future.
The reality is that traditional vehicles will not be eliminated overnight and, for the foreseeable future, traditional vehicles will need to co-exist with connected vehicles and self-driving cars in a mixed traffic flow. Simply speaking, most of the transport-related issues we are facing everyday will not automatically disappear as the result of CAV.
So, when it comes to modelling traffic with all types of cars, we have to use realistic assumptions and scenarios to more accurately understand the issues and better predict the benefits.
Which of your achievements are you most proud about?
Five years ago, the starting point in research related to self-driving cars was the assumption that we’ll have 100% autonomous vehicles. I think very few researchers were paying attention then to the transitional period. I was one of the few researchers who realised that the transitional period is the most difficult part of the journey to fully automated vehicles, which inspired my DECRA project.
Another thing that might be worth mentioning here is, many researchers in transport engineering have been modelling road traffic purely from the engineering perspective, and totally ignored or didn’t pay enough attention to human factors, that is, road users’ role. Since my PhD study, I, along with with my PhD students, have been focusing on integrating human factors into traffic flow models to make them more realistic. It seems that now researchers in my discipline are paying more attention to human factors. My research activities and recent publications may have contributed a little bit to this trend.