WHO WORKS HERE

People Staff Interview

• Started with the "Konnichiwa Mycon"(Hello My PC) comics)

  When I was in 5th or 6th grade, I came across a comic book called "Konnichiwa Mycon" by Mitsuru Sugaya. It described programming from the very basics, and it introduced how to make a mathematics game. It got me thinking, "I might be able to make games".

  It happened that near my school there was a public children's hall, and I found a PC-8001mkII in their administrative office. So I asked the people there if I could play with it, and they let me. There was no one else using it and I had it all to myself, so every day after school I would go there, and even on Saturdays and Sundays I would bring my own lunch and played with the PC all day long. I bought a book for BASIC and I studied it, but back then I still couldn't make a game that ran properly.

• Joined an AI Research Lab in College

  In Jr. High I got my parents to buy me an MSX, and I made games to submit to "BASIC Magazine" and "MSX Magazine". Some of them were published and I received awards for them. After entering the Engineering department of a University in Tokyo, I started part time work at an arcade in Shinjuku, were I would see the latest games. This was when games such as "Ridge Racer" and "Virtua Fighter 2" were coming out.

  I picked an AI research lab in college just from the fact that I would be able to work with computers; and they had a Silicon Graphics Indy, and for my graduation thesis I wrote a program for editing hair. At the time I wasn't interested in the AI itself as much, and I was more interested in the graphics for the humanoid intelligent agent, that was being developed as an interface.

• Made responsible for arcade games in a game development company

  My feelings for wanting to make games had continued since Jr. High, and I decided to join a game development company after graduating. After joining, I worked on an arcade snowboarding game and a gunshooting game, and I also did ports from arcade versions to the PlayStation®2.

  Thereafter I volunteered to join a team starting a new PC game, but unfortunately that project never saw the light of day and was terminated. Just around that time, a former colleague who had joined polyphony invited me, and I also changed jobs to join Polyphony.

• The important things in a race game AI

  Because the last job I had at my previous company was a route searching AI, I was responsible for AI from the start at Polyphony. I was a bit worried at first, because I liked racing games as a user, but had never been involved in its development, and I wasn't too knowledgeable about cars; though I had a driver's license I seldom used it.

  But running an enemy car with the AI in Gran Turismo meant having them drive without fail, according to the car's dynamics, so it was more important that I understood basic physics. In other race games, the driving line of the enemy cars are often predetermined and car specific dynamics are ignored, but Gran Turismo always runs the cars with AI's, based on the combination of the cars and tracks. But of course the competition in a race cannot be represented just in calculations, so I asked more knowledgeable people in the company to learn from them.

  

• The Character of Polyphony Digital

  One thing that left an impression on me when I joined, was that everyone was making Gran Turismo. In my previous company, the members were split by projects, and there was this atmosphere like other projects were someone else's problem, and once one thing was complete, they would go onto the next project.

  At Polyphony, everyone is continually working on Gran Turismo, so even in car physics there is a long accumulated history, and they've built something that no one else could ever copy. The other thing is that there is very little restriction, such as something having to be done a certain way, and it is easy to work because of this.

• Regarding the recruitment for an AI Engineer

  Since I joined, I've been doing what I can to improve the AI, but with the combination of new car physics and new tracks, the situations that need to be handled is increasing greatly, and there is not enough hands to handle it all. This also means that AI development by human hands is reaching its limit, and in the future, we will need to utilize machine learning technology.

  In recent years, a machine learning method called deep learning has emerged, and there is starting to be a true utility value to AI's. This is a field that is definitely going to grow and with lots of room to grow, so I believe you will be able to do many things that you want to accomplish as an engineer if you start now.