Hi. Hi.

Please start by introducing your professional journey in academia.

Hi, very good morning.

Uh, myself.

I started my career in the year of 2002.

Thanks

From 2002 to 2000.

I was working as assistant professor from BB Raju Institute of Technology in Nanjapur then afterwards.

Uh, I did my masters degree from Jain to Hyderabad.

And then I will promote it from the year of 2008 to till until my PhD as I have continued as a Associate Professor in Department Office, BB Raju Institute of Technology.

From digital system and computer electronics.

Then afterwards I received my PhD in year of 2016.

Awarded on 24th March 2016.

My area of research is uh, while pursuing off my PhD is the VLSI technology and robotics. Once we developed FPGA accelerators for the robotics path planning algorithms and a different robotic applications, that was my main core area. So in my.

Supervisor from Jain to Hyderabad. I did my PhD once again means from the Jane to Hyderabad in 2009 to 16. It's a part time program.

Then after that I I have promoted as a professor in 2016, then in a year of in academics, once in a year of 2020, I have I got opportunity in 1920 time opportunity to take a leadership role as a associate Dean R&D.

So my achievements as associated in R&D.

June Good number of publications from the.

Understood,

Faculty and also myself and I have organized a lot of FBPS and there was a culture with me to attend a lot of conferences. I to be conferences and skill to write the journals in this particular one.

And specifically after that I got opportunity to go towards Singapore as a for the post doc they'll in the year 2022 I have received as a initially as a visiting research fellow. Then from the 20/2/2022, August to May 2023, I was there as a visiting research fellow.

I have worked under the Professors Thambi Palace, Srikantan, who is a uh, past chair of the Dean of the UH.

SCSC means it is full of computer science and data center. So there I worked under him. Then afterwards again the request came from the professor to join for the research fellow. So I was rejoined into the research fellow again back into Singapore.

Again, the month of July. So total I'm having of a approximately 3 years is exactly 2.9 years of.

2.8 years of my experience as a postdoc at NTU Singapore.

Well, this is my journey which I have compiled when you come towards a research part. I have added a good work.

Once after, uh, completing of my PhD, I was there with the embedded and VLSI and, uh, robotics knowledge by the time of my PhD 2016. And, uh, in the meanwhile, after submission of my thesis, I got opportunity to look into the one of the research proposal that is a Satyam science and, uh, it's a, is a science, very yoga and, uh, yoga and meditation. So how was the first person?

India to get as a technical background with a technical background achieve the uh research funding from the Satyam scheme, I have developed a embedded based yoga system.

Which is a part of the computer vision techniques which are we have incorporated along with my JRF. So we developed.

Posture analysis of a weather in yoga postures effectively we are doing of a posters or not.

So that analysis I did, that is a job number we did and next job number 2, we read that.

Uh, the Porsches, because the observatory thing is after we, we met a lot of interactions with the, uh, people who are there in.

Uh, Vivekananda, hi.

Hello.

Yes.

Yeah.

Shall I continue?

Yes.

OK, yes, this is my background. I have developed of a uh, yoga sciences, uh, computer vision techniques for the yoga sciences by that, uh, we got two I triple E conference papers from that. Then afterwards, uh, till I was progressing recently also we developed a few of the subject knowledge that will be following as I anthropometric. Developments for yoga sciences.

Let's

Because there was a lot of approaches we are having among that approaches. The first job we what what I did exactly after a capturing of a human postures, we compile that using of a Yolo models by using of a Yolo models.

We, we, my, uh, my past experience, which I worked in, in Singapore, uh, total on a ORB.

OK.

SLAM for the computer vision with a robotics we're using of a VLSI reasons it's FPGA accelerators we have developed there. So I had a good contributions for the Dyson company. Then the with that background knowledge after reaching I came recently to India in a year of 2024, July 1st that first week. So from there again I have started off developing of a.

Computer vision for the human postures. So the anthropometric developments are happening now. So we are in a close to compile that details with the deep learning approaches. So this is a one of the area which I am.

UH-1 version of the.

Developing now the second part of the my contribution is a smart shipping approaches which we are doing so.

When I was there in Singapore, I was there was a challenge observed. The Malaysians are over obesity. So they're they're facing a lot of economical challenges with the over obesity. So their expenses are more towards this part. Then the observatory thing is why don't we give SMS solutions for this particular 1 so that we given a solution for the.

We start our providing of A2 solutions. One of that one is in a phase. One which I have developed till now is using of your media pipe and also the computer vision techniques to evaluate.

Smart shipping approaches by that our hypothesis is the proper chaving is decreasing human right gradually not allow servants so that that approaches also we did that the second aspect from my end and the third, uh, now.

Uh, I'm developing of a, one more area that we are calling as a, uh, we are developing of a VLSI computer vision and also robotics in a format of EMR means autonomous mobile robots, which is useful for the, the today's, uh, top end adverb or either a Rockwell Automation. All these robotics industries must require with the edge computing techniques.

Area of background which I'm doing of a research, yes, along with that.

And methods. So this is a my.

Uh, Observatory, uh, I have published a round of a 12 journals and another 4 journals are in a pipeline. So they are in a review. The very recent one is 1 more. I got totally 5 funding projects among that means in year of 2016 to 22, I got A5 funding projects among that five funding projects just now I said one, one was Satyam, the other one was regarding of a.

From Sarab is here early career research award also I'm awarding in year 2017 by that I got a funding around of 50,00,000 of Indian rupees at that time. So I have developed a human robots.

Sorry, uh, robo assistive robots, which will support the elderly people. That was a thought process. Now we are enhanced that robotics into the, the other modeling, uh, part of developing office. Certain things like estimating of a human getting using of a computer vision, which again, uh, resonate and apply the signal to the robot based on a human gating robot will move forward.

So these are all these part part of that one. We submitted one work for the IEEE neural networks which we are representing of a human gating when robot and a human was walking together at incline and a decline positions and a straightened way and turns. A lot of kinematics challenges are there. We are working, we are working, we are working with that. But that was the one area. Now the we have submitted our work it is.

The review and the other work which we have developed as a post processing. So the soft NMS which again occlusive scenarios, the FPGA accelerators are not yet available. So which is useful because NMS having of its own limitation. So it picks after a only one bonding box after the IOU based on the confidence score. Now we came over the range of the bonding boxes among that to eliminate.

These occlusion scenarios so that we preferred a soft NMS which is my password which I worked in Indio Singapore and the other part is so developed. A HH BST means the Hamming and hierarchical model of the binary search tree for the future matching method. This is a again which is deployed in the Dyson for a autonomous vehicle for a future matching of images and things because lot of the.

Driving and returning towards the memories and out was please I have used a Zing ZC 104 and NTU and that was my skill set and I'm good enough in a artificial intelligence and a computer vision and the along with that I have.

Having of a these assistive technology, robot developments and so on. So these are my background.

Thank you for providing this detailed overview, Professor. Allow me a moment to organize my thoughts, and we'll proceed.

Yeah.

Hello.

OK.

Yeah, my teaching experience. I have started my career in a year of 2002 as a assistant professor in AB in college like a doctor, VRK Engineering College for one year, then afterwards in 22,003 to 2004 I have worked as in in a Narayana College and Nellore district Guddur for one year, then afterwards from year of 2004 June.

2008 I was there as an assistant professor, then afterwards I was there as a.

Uh as in the same organization I have promoted as Associate professor in the year of 2008 to UH-16 as Associate Professor. Again after my PhD I have promoted as a professor.

In 2016 March until that I'm a professor and also ratified professor from a affiliated university JNU Hyderabad. In Hyderabad we are having a very less ratified professors in the department office among them at Max in a year when I have ratified in 2022.

Among Ten 1% and this is my. Then of course I was there in a post doc. I went to postdoc there.

And Singapore for three years, then again back. Rejoined as a professor in BV Raju. Instructor comes in the year of 2025, July. This is about my teaching.

Hello.

Yes, Professor. Please continue.

Yeah, Thank you. I have a provided teaching experience.

Hello.

Yes, Professor. I am here. Please let me know if you'd like to share anything further or if you're ready to proceed with the next question.

Yeah, please. Next question.

Professor, you've conducted research on FPGA-based accelerators for feature matching and occlusion handling. Could you summarize these projects and explain how you integrate your research into your teaching methods?

Right, thank you for that comment. Was the part of that my story uh which I have worked for the?

Uh, in generally the teaching we use up to the level of certain existence of the, uh, developing of a digital design and deploying that one into the Fpgas and that will do an undergraduation or post graduation or PhD for a some exemplary things like comment was the part of the story is that.

Point #1 If we, we, we are having of A1 particular thing which I have worked as a FPGA accelerator for a soft NMS before Windows is uh, the soft NMS is uh, soft NMS is a part of the post processing level of the Yolo model. Means in a Yolo model, the level 1 we'll capture of a images from that images, we'll use a conversion neural networks from that neural.

Works with captures the data and when we compute that data and then afterwards means all this convolutional neural networks are we are going to position at PS model means in ZC 1:04 we position it in that that content into the PS. Then the post processing which we developed of the.

Soft NMS is the architectures. Again, accelerators are deployed in a PL. so we establish the communication between.

The PS and PL. based on the image which we are going to use for the occlusion scenarios, we compute that occlusion scenarios. How we come we compile that occlusion scenario means. Just now I mentioned as a IOU intersection over union. Among the intersection over union, we have integrated of a confluence method because again there is a challenge the general IOU, general IOU and BIOU CIO and all the different types of.

How you start there we brought something as a confluence IOU in place of general IOU so that IOU is providing of a best solutions than the general IOU. And this is a part of the which we are calling as a confidence score of my architecture. So it enables it enables my software Ms. and provide that the range of my bonding boxes among the range of bonding boxes which are very close associate in.

Conclusion scenario, we consider that occlusion scenario effectively and take a decision the other end by either robot or autonomous vehicle that to project that one. If these occlusion scenario is not considered effectively at the post processing model, lot of accidents are happening in the traffic. So that was the one of the NMS drawback. So we are not. In the NMS is a challenge today in the market, so.

Soft NMS was provided.

Daisy, The research of the FPGA accelerators for the occlusion scenarios.

Right that this how this is deployed to the teaching methods means when it comes to the teaching method to do any research work, we required this background of how to use the PS and how to use the PL. and when the control algorithms are required to interference between the PL. and PS. So at the time of execution there was a one of the. Teaching learning is a bigger thing for the students for the.

Well, they work for the real time works.

Yeah, this is my solution Barry.

Thank you, Professor. You’ve thoroughly explained your research and its connection to teaching, particularly in blending advanced FPGA concepts with real-world applications like occlusion scenarios. Could you elaborate further on how this approach supports students in developing practical problem-solving skills and prepares them for industry or academic roles?

Right comment was this, uh, we worked at after the extent of the soft NMS, but.

There was still a lot of hypothesis are leftover means the bigger area like we are having of a IOU. Just now I said as a intersection over unions 1 aspect among that there was a certain things are there still challenges as a GI OU means generic IOU which is which is the bonding boxes are away based on the improper classification also that challenges are available so to map from.

That particular 1 from 1 end to other one, we required some understanding of GI OU and not only this, but 11 CIO's are required, DIOS are required. And in the other end of the NMS method, we've seen of a just only the uh up to the part of the NMS and soft NMS which we progressed at the soft NMS for occlusion scenario. Yes, we also required adaptive NMS method in future.

For this type of way, uh, real time solutions, so, but coming towards the, the students when their works for their real time analysis, most of the robotics and autonomous vehicles are required computational computing of a things and edge computing because we, we are dependent more on a cloud today, but cloud is a challenge as a proportionate other end versus GPU. Yes, if you use a GPU, these scenarios will.

Out but The thing is your cover power consumption at what rate you are losing of your temperatures and everything is again computed into the to the autonomous vehicle. So the edge computation solutions will provide a better solutions is A1 solution for the students who can work in this lines. And second part is not only this yes we can scale down this one in place of for the small robotic applications like a parking of a robot where we are having a.

Scenario In a clustered scenario, the robot autonomous AMR should understand and take it right decision with the computer vision and position for a parking alright navigate in that area. All these are all the exemplary things which are useful for the students to use these occlusion scenarios versus.

FPGA accelerators. This is from my end.

Thank you, Professor. Your explanation clearly highlights the connection between your advanced research and its application in teaching. You’ve shown how your work prepares students to tackle real-world challenges in robotics, autonomous vehicles, and edge computing by exposing them to state-of-the-art methodologies like soft NMS and IOU computations. Moving forward, can you provide an example of a specific challenge students have faced when working on these advanced topics, and how you guided them to overcome it?

Yeah, the major challenge when we face, we design our uh, architectures like uh, our own flexibility, but uh, the FPGA accelerator doesn't work as per our flexibility lot of because you are dependent on the DSP slides and Luts and a lot of the constraints you are, you are device area is much important so.

And understand the computer vision and come back towards your RTL and recorrect it and validate it with the system very lock. And because all the verification should be done at the higher levels, not only this particular 1 you you are much aware of the AXI protocols and all these things. So the communication protocols are much required in this designs. When we the challenge what I have faced remains.

Your architecture again your RTL should be perfect.

Uh, when the first, uh, HSBSD, the feature matching algorithm, when I'm developing that, uh, redundantly, I went towards the, the biggest architecture. So that was not fit for me for the implementation in FPG. So again, we optimized and came out with the solutions for that, uh, some Hamming method and came into a solution Binary means the bit sizes, all these compilations I did from my end, whether we are able to.

Resolve that and provide a two days and that is a success story from our end.

Thank you for sharing this, Professor. It’s evident that you have tackled and overcome significant engineering challenges, particularly in optimizing FPGA architecture and aligning it with computational requirements. Your ability to translate these technical challenges into learning experiences for your students is commendable. This kind of mentorship undoubtedly equips them with the skills needed to address complex real-world problems. Let me wrap up this section. Thank you for your detailed insights on your research and how it informs your teaching. Let us proceed.

Thank you.

OK, uh, comment was in my future, uh, the present I'm having of a totally.

You're welcome, Professor. Let's continue. Could you outline your three-year research roadmap and explain how you plan to align it with the department's goals?

Three areas of a product development means after the Singapore international exposure to me, I got a picture. Hello you there.

Hello.

Uh, before Windows Singapore, my thought was run behind the two things are there. One was rewarding of the technology development from the area of FPGA accelerators, so it is given opportunity to go Singapore. The second part was a much connecting towards the, the human Health Sciences. So musculoskeleton analysis, yoga portions, all these things. So now I'm working for my part is one part is I want to develop a product like a smart shipping mechanism using.

Yeah, uh, after international exposure, I, I, I got a clear picture about me that they.

The computer vision techniques, that was it, the preliminary phase one and phase two of that was we are going to come some variable, uh, earbuds that connects with the analyze of a my, uh, achieving mechanism of this. I have visited for this two, uh, uh, institutions, dental sciences, which are a top most in India, like under the NIF ranking, the Savita Dental College and Vishnu Dental College. I had a collaboration with them.

Understanding of the concept and getting whether my hypothesis correct or not, whether the how we are going to get this part. So this is a phase one which I'm working with the software development and second one is this hardware development. Just now I said is a earbud for the analysis. That's a phase two of the work. And third phase after a successful of this product development and based on the market analysis, I wish to go something as a semiconductor.

Then now the Indian semiconductor is also providing some chip designs and this. So the in future we'll integrate means we'll come out with a a effective way of SMS chip model that was a phase three but not immediate based on our market and a team. So we'll look into that part that is a regarding of a smart chiming approach which I'm considering here at the level 1.

Second part was, uh, proportionately, I will publish all the, uh, papers for the these uh, hypothesis which are not affects my product. So the subsidized materials I will publish for the I AAA transactions and so on. Coming to the second one is I'm more into the second is a yoga census with a musculoskeleton analysis of the human body. Now we are analyzing something calling as a region of motion.

Which we are calling as a negating techniques which says that the every individual is not having of a equal limbs. The lower limbs that is a my recent observations that based on their limbs and a different alignment of their limbs, their gatings are different. So for that again the deep learning is required over. So we are processing of their deep learning with the computer vision with a again FPGA accelerators for that one.

This will.

That was the one area we are working the second part and 3rd part we are working for the another one is a yoga Porsches with the photo anthrometric methods. Now the third one which I I'm working with a photoanthrometric works. Are they based on my?

Useful for the people to have effective walk and by that.

Photoanthromatic computational methods.

How human is getting getting is very connected and also is a Porsche's Are there Porsches are performing of a different yoga Porsche's. OK that was the third part and 4th part which is better and better for me from long time. That is the all the computer vision again back towards the robotics which are much required for the adverb and a Rockwell Automation. Yes, recently we visited A VLSID conference. There we got opportunity to connect connect.

The Rockwell Automation team so shortly I'm going to connect with them more to get progress into this type of lines means they if you look my stretch there was the two things are there computer vision is 1 solution and 2nd solution is a embodied division and a third one is a FPGAS accelerators. So the total complete system level designs and a precisely which is useful for the human being for the future.

So, uh, outcomes I'm going to get in a three ways. One was part of getting off a publications that will come. The second part is getting off a funding, uh, for this. Again, if you require from the institution or a from the, uh, government, because I have already experienced and did have a 2 findings as a principal investigator and other three fundings as a Co principal investigator. That was another area so.

Uh, the other things right, fine. These are my feature I ideation. I want to come out with a product development with the span of five years. I want to bring at least a two to three products into the market afterwards. I'm working for that now.

There's a mile learning from my end.

Yeah, this is from my.

Thank you, Professor, for sharing that comprehensive roadmap. Let me probe a bit further on this. How do you plan to mentor junior students to ensure a consistent pipeline of high-quality publications and researchers within this ambitious framework?

Time is up! Ending the interview now.