In a recent post I wrote about this project during the eNTERFACE '08 workshop, where I spent the whole month of August with a multidisciplinary and multicultural team. Just to give you a fast overview, the project was about the fusion of 2 different modalities in order to help people on their every day life. The two modalities were speech and image, detected by a speech recognition tool and an image processing tool respectively. The fusion mechanism combines speech and movements to predict what people are planning to do in a certain scenario and the computer should help them to complete their tasks as fast as possible. For a scientific project and considering many challenges and limitations, actually it was a successful project. We are all happy with the results.
I'm presenting the evolution of the project in a middle term presentation.
So, what could be so challenging in a project like this? This is a list of challenges we had and I would like to share it with you and with future eNTERFACE attendees:
- multidisciplinary team: our team was composed of 5 people from 5 different specialties, which were: linguistics, system engineering, speech recognition, image processing and computational semantics. All these skills were important to achieve good results, but most of the time people worked alone, each one in his/her specialty, meeting only when an integration is needed or when it is necessary to know the status of the project. In terms of scientific results, it was good, but in terms of problem solving it is more efficient to work in pairs. We defined a very challenging target and a multidisciplinary team was good, but if you have a very specific target to achieve, consider more than one specialist per discipline.
- multicultural team: Since eNTERFACE is an international workshop, people come from different parts of the world, and there is a big probability to have many nationalities, with different cultures. Our case was extreme ;) . 5 members, 5 cultures, which were: Brazilian, Russian, Swedish, Mexican and Chinese. 4 continents in a room, trying to conciliate their habits, ways to express themselves, ways to think about the same problem and so on. With a multicultural team, you never know how to react in some situations, and the best thing to do is try to be as much transparent as possible. Without transparency, a cloud of assumptions stops on your head and the project might suffer a big impact.
- time constraints: You have a big scientific challenge in our hands and you have to achieve it in less than one month (19 days actually :O), otherwise, you won't have something original to show at the end, which might be a little bit frustrating. It implies in almost no time for tourism (I had just one opportunity and I wasted it in EuroDisney :( ), many nights sleeping very late, and some level of indiscipline with food and self care. But, don't worry! When everybody is in the same situation as yours, actually you don't feel bad at all, but challenged.
- dynamic scope: scientific projects have, since their conception, a very weak scope. Don't try to emphasize the scope because it is a natural limitation for creativity. Every time you talk with somebody there or attend an invited speaker session, many ideas start to appear and you can not just say no for them because you actually like them. So, what? A good approach is to think the inverse. Instead of limiting the scope, define milestones describing what should run at that stage and not how it will run actually. Good researchers need freedom of choice and thinking, but they also want to know what they have to achieve at the end.
Our eNTERFACE team (from the left to the right): David Gomez (Mexican), Olga Vybornova (Russian), Daniel Neiberg (Swedish), Me (Brazilian) and Shen Ao (Chinese).