Hello, this is Sobue of moat Fujimoto laboratory Master’s 1 year. Currently, I am doing a research internship at NASA’s Jet Propulsion Laboratory. In this article, I would like to introduce a little about the experience of “research internship abroad in Corona”, the background to the decision to participate, and the current activities. I’m currently in a city called Pasadena near Los Angeles. COVID19 is currently remote work, but it is expected that on-site activities will be eased soon.
First of all, the background to my participation was that in the winter of two years ago, a senior doctoral student in the laboratory at that time participated in the same research internship program for three months (currently graduated). Even after returning to Japan, he summarized his achievements at the time of participation in a dissertation in parallel with his doctoral dissertation, and was in regular contact with the local team. Meanwhile, last fall, the team happened to ask me, “I’m looking for a student intern who seems to be suitable for this project, but is there any acquaintance like that?”, So I introduced me via that senior. I had the opportunity to do it. In Europe and the United States, in such cases, your academic background, work history, research content, achievements, skills, etc.
It is common to send a CV (Resume, Curriculum Vitae) that describes the website etc. to the other party. I asked my senior to send me my CV, and about a month later, he said, “The research theme and skills match the project, so if you can get a scholarship to cover the length of your stay, please join us. I want you to do it. I will issue a letter of recommendation for scholarship acquisition from here. ” As a result, I was selected for the “University of Tokyo Toyota Study Abroad Scholarship for Advanced Artificial Intelligence Human Resources Development”, so I cleared the conditions and successfully participated. As for why I was asked to get a scholarship, I think the reason is that I am “screening” for those abilities. In fact, when applying for a scholarship, I created a document that describes the current research content and the research theme that I plan to conduct during my participation. Researchers also go there every day. In Japan as well, when applying for a researcher at the Japan Society for the Promotion of Science, such documents may be required.
The adoption of this scholarship was decided at the end of March last year, but unfortunately the corona scholarship began to spread all over the world. Even at Japanese universities, what happened with lockdown and whether face-to-face lectures could be continued, so I asked the other party to change to a plan to conduct a research intern for one year from October, and once finished the interim presentation of M2. I decided to take a leave of absence after that. I participated in weekly meetings from around August, and started teleworking in Japan from October.
Next is the current activity. The project I’m involved in is the “DARPA Subterranean (SubT) Challenge”, a type of robot contest sponsored by the US Department of Defense. This involves throwing multiple robots into an underground environment such as a cave, tunnel, or inside a nuclear power plant, and within the time limit, find a doll-like “artifact” that imitates an injured person, and find the number and estimated position of those “artifacts”. It’s about competing for accuracy scores. There are two types, an online version and a real version. Universities and research institutes in the United States and abroad form a team to develop robots and their autonomous systems. The team I am on is made up of JPL staff and student internships, and the number of people is less than 30.
(This is a team logo photo)
Since the actual field is not known in advance and the communication environment is not given, the required technical elements are as follows: (1) Build a network while dropping the communication antenna (2) Each robot estimates its own position with the start position as the origin (3) Map of the environment ④ Determine the internal search policy and make decisions and control ⑤ Find the position of the artifact by image recognition (and send it to the entrance). Since these are performed by multiple robots at the same time, the performance of the software determines almost everything in spite of the robot competition.
Before joining in October, I was assigned to the team that decides (4) internal search policy and makes decisions and controls. In a familiar example, this team is developing software that “makes a route map while digging a tunnel and visits many stations while expanding the route map as much as possible within the time limit” in a given environment. The problem of making such decisions in an unknown environment is called the “partially observable Markov decision process”, and I started by improving the software in the “drilling tunnels and making route maps” part.
Fortunately, the work was software-centric, so since October I have been attending the entire meeting twice a week while communicating with the team’s slack, and I have been doing research internships as such. At zoom and Webex meetings, it can be difficult to convey things like nuances that are difficult to convey unless you are face-to-face. However, the experience of working together as a team to develop one huge system or software is a lot of fun. In addition, since the program is simulation-based on a scale and the results of the field test are developed once a month, we are able to develop with the actual thing in mind.
At present, I think that many students are living the research life that many students had expected due to the coronavirus all over the world. Some students may have decided to take a leave of absence for a different reason than I do. That is why I hope that I will be able to find a field that interests me, work hard on my studies and research, and when this situation is over, I will be able to look back and spend a meaningful time, and I hope that I will be able to use that time.
It’s also painful that you can’t take pictures that look good on these blogs online (bitter smile).
This page was automatically translated into English.
Please note that the translation system may not necessarily be accurate.
Thank you in advance for your kind understanding.