Master Interviews in 2022

3D pedestrian route choices and my research days


£with laboratory members@at summer school

Eiji: Where are you from?  

Shen: I'm from Zhejiang Province. It's a city with a long history, 
and my university is Chang'an University in Xi'an. I majored in 
Transportation Science, and when I was an undergraduate, I worked 
on pedestrian-vehicle interaction, modeling and simulating 
pedestriansf decision process using Decision Field, a psychological
 theory, and published a paper on it. This experience made me find 
that the pedestrian is an interesting object of study, as their 
routes are flexible, often unconstrained by traffic facilities, 
and their decisions are often dynamic, easily be influenced by 
the environment.

Eiji: Why did you come to Japan?  

Shen: When I first started in the field of pedestrian traffic, 
I read some papers by Japanese scholars, and some of them are 
(maybe were) working at the University of Tokyo. Also, Tokyo is 
a charming city: itfs well developed with high-tech companies 
and clean streets; and has a traditional culture similar to 
China, as well as many interesting subcultures, such as games 
and animation.

Eiji: While there are still many foreign students who cannot 
enter Japan, Mohammed-san came to Japan once, but now he is 
continuing his studies in Lahore.

Shen: Yes, Ifm lucky to have the opportunity to study in Japan 
during COVID-19. It was very difficult at first, I wasnft sure 
when the COVID-19 would ease off and when the Japanese Government 
would open for foreign students. Although the university and your 
lab had provided me with a nice environment for distance learning 
and research as possible, I was really worried that I might have 
to spend the whole two years in China and complete my master's 
degree online. If so, Ifll miss out on so many interesting 
experiences and lose some opportunities for personal growth. 
Because in my opinion studying abroad is not only about studying 
and research, it is also important to experience the local life 
and culture and to interact with people from different backgrounds. 
I finally arrived in Japan in November of my M1. I stayed in a 
hotel for a week after I arrived... The window of the hotel is 
my first impression of Japan. (laughs).
Eiji: Oh, I see. So that's how it was.
Shen: When I came to the University of Tokyo for the first time, 
Koseki-san showed me around the Hongo campus, itfs very beautiful! 
And the lab members gave me a welcome sushi party, which made me 
feel at home. The lectures were different from professor to 
professor and the way of examinations was also different in 
COVID-19. Due to the risk of infection in organizing offline 
exams, most courses were graded in the form of reports, which 
allowed me to put into practice what I had learnt in class in 
small projects. It was very interesting, but anyway I had the 
impression that time passed very quickly because I was learning 
new knowledge and skills online. For my master's research, I 
worked on a pedestrian model in 3D urban space using 
multi-sensors, as it was close to the theme I had done in China.
Eiji: How difficult was your research?
Shen: In my research, we observe pedestrian travel behavior 
with multi-sensors and formulate observation equations.  At 
the same time, we estimate the parameters of the behavioural 
model and conduct simulations. Data processing involving some 
machine learning approaches is challenging, but the most 
difficult part is the optimization method. I spent a lot of 
time learning discrete choice models from scratch, which is 
quite different from the model I used in my previous lab in 
Eiji: How did you learn it?
Shen: The start-up seminar led me to understand the research 
theories used in this lab. Also, the reading seminar organized 
by Arai and ogawa in the summer vacation of M1 helped me a lot. 
As for the Recursive Logit model, I read the Fosgeraufs paper 
and related papers published by your lab. The papers by Oyama 
and Eiji are easy to read, and the archives of the presentations 
at our theoretical debates were very helpful.  Also, I am very 
grateful to Kobayashi and suga for answering my questions on 
RL model.
Eiji: It's nice to have different mathematical methodologies 
running together in our lab, so we can learn from each other 
and stimulate each other's research. How was your coding 
Shen: When I was in China, I mainly used off-the-shelf machine 
learning algorithm packages to process images and get 
trajectories, and the programming requirements were limited to 
the pre-processing of the data. But in this lab, we wrote the 
code by ourselves. It was challenging but interesting. To 
complete my masterfs thesis I have to try to write some 
algorithms by my own.
Eiji: What was interesting?

Shen: I liked the part about link switching more than machine 
learning, I didnft finish it until the last week before the DDL 
of master thesis. I wrote the code for switching in the 
pre-processing of route data from scratch by myself, looking 
at Oyama-san's algorithm, and I was really happy and relieved 
to have done that.

Eiji: What do you think of the potential of machine learning? 
In the ACT-X project, where I am on the committee, and at the 
Next Generation AI Center, is a programming exercise every 
year, and there is a big difference between those who can 
achieve a completely new coupling of machine learning and 
theiroxn topic, and those who can only come up with a mere 
comparative methodology with an exam-like approach.

Shen: Machine learning is certainly in vogue these days. A lot 
of researchers are working on it. Then, the key is how to 
improve machine learning, how to adapt it to different 
research scenarios and data structures. I think preprocessing 
is more important. I felt the same way in my master's research.

Eiji: I see. What is your impression of Japan? You're in the 
minority in Japan, and you've had a hard time?

Shen: Evangelion. When I was a child and I saw Eva for the 
first time, it was very different from normal anime. As a 
child I was scared. But it was one of the first images I had 
of Japan. In both the anime and the film, Shinji, the main 
character, is a controversial character who is not a typical 
hero: he does not have a strong will and is even cowardly, 
always running away from his responsibilities. But that's 
what makes him so real, he is our friend, he is close to us.

Eiji:  Even if you want to run away from reality, whether it 
is your thesis or job hunting, it is human nature that you 
cannot. It is a feeling that we all have.

Shen: I think so. I think there is a heroism in Shinji. Hefs 
running away from himself, running away from responsibility, 
and he led to Third Impact, and that was the mistake he made. 
But in the last choice, he stood up to the responsibility he 
had feared, to the irretrievable and dire consequences he 
had caused, and that was very valuable. This is heroism for 
the common man.

Eiji:  In my own experience, studying abroad is like resetting 
the past in a way, isn't it? There are things you can only 
understand when you feel as if the past has been cut off. But 
Eva's story made me realize that in the end, we are connected 
to our daily lives and reality. We can never restart like in 
a game. Eva ends with a scene from Anno's hometown, Ube. What 
was it about the Japanese landscape that made such an impression 
on you?

Shen: My impression of Japan is in COVID-19, so it's mostly 
Tokyo. I like jogging since I came to Tokyo, itfs a safe way to 
enjoy the scene of streets of Tokyo, and because of my research 
theme, I also enjoyed seeing people running and walking around 
the streets of Tokyo. I used to run about 5km from Hongo campus 
every week. (laughs). Taketomi Island was also beautiful. I'm 
glad we could travel there together. On the island, we rode in 
an ox-cart and sang folk songs together. I was also impressed 
by the starry sky we saw on the seaside bridge at night.

Eiji: I'm glad I went there at that time.
Shen: When I graduate, I want to go on a trip and see the cherry 
blossoms in Kyoto. So I'm going to stay in Japan until April. 
Laughs: After that, I will work on research into automated 
driving. I want to continue my research on traffic, on the 
interaction between pedestrians and vehicles.

Eiji: Research on interaction is just like the world of Eva, 
where emotions and reactions change as the environment changes. 
I think it would be fun to do research that goes beyond mere 
image processing. Recursive itself, which involves mutual 
decision-making, also changes, so it could develop into a 
problem of game, equilibrium and estimation of behaviour, which 
could lead to doctoral research. Finally, if there is anything 
you would like to say, please do so.

Shen: My first impression of you was a YouTube video talking 
about the Olympics, in which you were a typical engineering 
professor, serious, erudite, and intoxicated with research. 
But when I came to Japan and met you, I saw some more aspects 
of you in real life and seminars. You like comics, would 
discuss the Olympic table tennis competition with me, would 
talk about the responsibilities of the young generation and 
the vision of the future after the seminar. Ifm glad to be 
supervised by an interesting and charming person. I had a 
hard time finishing my graduation thesis, the support of you 
and other lab members really encouraged me, thank you very much.