Carlos Gershenson, PhD, combines academic interests in complexity, artificial life, information, evolution, cognition, artificial societies, and philosophy with his role as Leader of the Self-organizing Systems Lab at the Universidad Nacional Autónoma de México. Durtti wants to find out about the solutions the lab is developing to increase the efficiency of urban mobility.
Tell us about the cost effective self-organizing traffic lights solution you and colleagues have developed, Carlos?
Traditionally, traffic light coordination tries to synchronize phases with the expected traffic speed, so that vehicles don’t stop. However, traffic changes constantly, and it is common that vehicles cannot move “as fast as expected”, and this synchronization scheme might even increase waiting times.
In contrast, we propose the use of sensors and simple rules so that each traffic light can adapt to the current situation, giving preference to the street with highest demand. This causes that places with few vehicles have to wait more, but have the chance to gather more vehicles, promoting the formation of platoons. Once platoons are of a certain size, they can trigger a green light before reaching an intersection, so they do not need to stop, unless there is another platoon crossing. Computer simulations have shown that this approach delivers optimal or close to optimal performance for all traffic densities.
Which 3 other aspects of society do you believe would most benefit ‘commercially’ from adopting self-organizing system applications in the next 3-5 years?
My lab has focussed on urban mobility, but most urban systems could benefit from the adaptation that you can get with self-organization, as urban systems “live” in ever-changing environments. Bureaucracies, markets, education, could also benefit from being coordinated through self-organization. But for the next 3-5 years, I believe that the highest impact could be achieved in behavior regulation through self-organization. In particular, there are plenty of opportunities related to health (how to spread healthy habits?), democracies (the ones we have are certainly obsolete), and well-being (leading a fulfilling life).
If Claude E. Shannon, the Founder of Information Theory was alive today, what one question would you have for him?
How would you relate information to matter and energy?
Big data often contains complex interdependent components. How would you go about measuring ‘complexity’ in order to add significant value to your overall analysis of the data?
There are dozens of complexity measures proposed in the literature. We have one based precisely on Information Theory, which basically reflects a balance between robustness and adaptability. However, it is important to note that even for the same measure, complexity can change depending at the scale at which it is observed. And even if you have a measure, the interpretation part is not trivial, i.e. if you get 0.89 it doesn’t tell you much unless you contextualize the value.
As humans, we all have good days and bad days! Has becoming a father changed the way you deal with obstacles in life? If so, how?
I think one of the wonderful things about having children is that they make us realize many of our positive and negative aspects, as they imitate almost everything people around them do. We want the best for them, but then we realize that the way of achieving that is learning how to get the best for us. So it is a great opportunity to develop patience, empathy, and compassion, which are useful not only with our families, but with everyone we interact with.
As a keen cyclist, how much safer do you think roads will be if and when level 5 autonomy of motorised vehicles is achieved? Will there be other safety issues for cyclists to consider then, in your opinion, that don’t currently exist?
About 95% of fatalities in car accidents are due to human errors, and most of those are attributable to the 4 D’s, i.e. driving while distracted, drowsy, drunk, or drugged. Already with current technology, autonomous vehicles reduce the risk of fatalities. So in principle cyclists should be safer. But then those are averages, one should always be alert, not only for vehicles. The risk I see is related more to political and economical decisions. If cities promote infrastructure of autonomous vehicles over others, i.e. restricting even more roads to cyclists and pedestrians, then in practice it would be against our right to mobility, as people would be kindly forced to use autonomous vehicles.
What is the most valuable lesson you have learnt from teaching others?
Knowledge can help us make better decisions in all aspects of our life. Thus, it is highly motivating to share knowledge for the benefit of all.
Finally, Carlos, what advice would you give to a class full of primary school children about how they can best prepare for the workplace of the future?
The world is changing fast and it will change even faster. So the usefulness of education has shifted from learning specific abilities to learning how to learn. The better we can adapt to the changing future, the more suited we will be for future workplaces. Interestingly, a key aspect of adaptation is creativity, which is currently not promoted in most educational systems but companies demand it more and more.
Carlos is a member of The Artificial Intelligence Group on LinkedIn.