Blog


Interview | Prof. Mohan M. Trivedi, UCSD

background

“Peer-to-peer interaction is extremely crucial”

DTU Times interviewed Prof. Mohan M. Trivedi, Electrical and Computer Engineering, University of California, San Diego, who was the Guest of Honour at the 1st International Conference on Signal Processing, VLSI and Communication Engineering (ICSPVCE - 2019) organised at DTU.

How have you seen the field of Computer Vision and Robotics evolve over the years?

" One of the most remarkable achievements was the incorporation of Machine Learning."

When you’re a part of the journey, you don’t notice much. But the key changes that have taken place are the resources that are required have increased multifold. When I was a graduate student doing object detection, we were looking at wildlife through aerial sensors and it would take us days to get the results. But today when the students perform such analyses, it gives vast results in no time. Prof. Raj Reddy, a pioneer in Robotics and Computer Vision, had a philosophy that the dissertation results weren’t accurate unless they worked on 4 different types of images. That time it was a rigid thing, but today we perform experiments on large datasets of images. One of the most remarkable achievements was the incorporation of Machine Learning. Earlier, we used predefined features, which needed specific knowledge of the data. Today we have data for every feature available for learning.

Collaboration between academia and industry is very important. In San Diego, as shown in your lecture, you’ve collaborated with industry giants. But in India, we seldom find such collaborations and the existing ones are too limited. What are your views on this?

I think everybody realises that we have the same objectives at a higher level. But we have different expertise and infrastructure. Unless we really put it all together, we cannot solve the problem and nobody wins. So once people have developed some trustworthy relationships, good things will start happening. The industry also is hiring students from universities, so they have a very good idea of happenings inside the universities. The task is the same but the difference in an industry is that there it is commercial and profit-making. The freedom to think freely might not be available in the industry. But if you know people with similar ideas and thinking, you’ll feel comfortable being and working with them.

It is often said that students need to be motivated to pursue research? Do you really think a motivating factor is needed, or this should come from within?

Most successful students I’ve seen have a high amount of self-motivation. Whenever I need to bring in new students, I tell them to interact with other students in the group because peer-to-peer interaction is very important. It might not be necessary for some theorist like Ramanujan, with high thinking skills, but this isn’t the case with us. We certainly want to emphasise that people are respectful of others ideas and help out each other. People understand that I might be helping you today, and you might help me tomorrow. People then enjoy working and collaborating.

What is the biggest challenge that the Robotics industry still faces even today?

"... works in real-life with continuous interaction, without humans feeling threatened"

Safety, Robustness, Generalisation and Human Centeredness. I believe that making one-of-a-kind is no big deal, but making it robust and able to work in a variety of situations as well as interact with humans is very important, and hasn’t been yet achieved. There are a lot of open issues left unresolved. A robot that works in an artificial environment is not needed; what’s required is one which works in real-life with continuous interaction, without humans feeling threatened or stressed.

What final message would you like to give to the students?

I am very impressed and glad to visit DTU. I really wish a large number of students from here contribute to the community of Robotics and Intelligent Systems. I’m sure you’ll be successful. I want to re-emphasise that particularly in this field you might end up building something life-saving. I wish you all the best!

Posted by Parangat Mittal

@thesciencestudent