A Q&A With Renowned Neuromorphic Chip Designer Chiara Bartolozzi


Every year for Women’s History Month, All About Circuits spotlights the contributions of distinguished women engineers worldwide. For this article, we interviewed Chiara Bartolozzi, a senior researcher and neuromorphic chip expert at the Italian Institute of Technology (IIT).

Since earning a degree in engineering from the University of Genova and a Ph.D. in neuroinformatics from ETH Zurich, Bartolozzi has led important research in neuromorphic engineering. She also helped design iCub, a toddler-sized humanoid robot developed at IIT that serves as a robotics testbed worldwide.

 

Chiara Bartolozzi with iCub, the humanoid robot developed at the Italian Institute of Technology (IIT). 
 

In this exclusive interview, Bartolozzi delves into her professional journey, touching on her greatest achievements and vision for the future of neuromorphic engineering.

 

 

What drew you to the field of engineering?

Dr. Chiara Bartolozzi: I was one of those high school students who was interested in all topics: literature, history, philosophy, mathematics, and physics. It was difficult for me to decide what to pursue for higher education. Ultimately, I chose biomedical engineering, which was an emerging field in my hometown at the time. 

Perhaps due to its link to medicine, biomedical engineering attracted more female students, particularly compared to mechanical, electronics, and informatics engineering. Drawn to the idea of helping people, I initially set out to design prosthetics that could heal or restore lost body functions. After attending a visual neuroscience course where we modeled computations performed in the visual cortex, I became interested in brain-inspired circuits.

 

What are your key research interests? How did these interests evolve over the years?

Bartolozzi: In my Ph.D. thesis, I developed a synapse-inspired chip inspired by the human visual system, particularly the selective attention mechanisms. Such a system helps us understand the location of something important within our visual field so we can orient our gaze and capture it in the high-resolution part of the retina (the fovea) while capturing its periphery in low resolution. When given visual input, the chip I introduced identifies the most interesting part of the stimuli, so you can point a camera at it. 

 

Schematic diagram of DPI

Schematic diagram of Bartolozzi’s novel VLSI synaptic circuit, the “diffpair integrator” (DPI), connected to test circuits that augment the synapse’s functionality. Image used courtesy of IEEE
 

I subsequently moved to IIT and started exploring the applications of the chip I developed in robotics, particularly on the iCub humanoid robot. Because the robotics community focusing on neuromorphic engineering is still relatively small, our progress has been rather slow. Currently, we are developing circuits for tactile sensing and algorithms to gather information from sensors.

 

How do you anticipate the field of neuromorphic computing evolving in the next few years?

Bartolozzi: The field is growing in many directions. On one side, there’s an ongoing effort to create platforms that can simulate many neurons. On the other, there’s the integration of in-memory computing circuits to miniaturize systems.

 

Chiara Bartolozzi with iCub

Chiara Bartolozzi with iCub, the humanoid robot developed at the Italian Institute of Technology (IIT). 

 

Another nice direction that is emerging is the concept of embodiment. When looking at how the brain solves problems, a lot of information comes from the fact that we have a body; we know how we move and how our body exists in space. In neuromorphic engineering, embodiment entails combining a lot of information gathered by different sensors while also accounting for the movements that generate this information. I would like the field to progress in this interesting direction, but if it does, it will take time.

 

Can you tell us what you most love about being an electrical engineer and what you find most challenging?

Bartolozzi: I am fascinated by working out how to use transistors beyond the digital domain. The advantage of using transistors in this way is that you have very low currents and very low power consumption. The underlying physics resembles that underpinning current within cell membranes. I am excited by what we can build with this approach, including very compact circuits that can replicate higher-level behaviors in robots.

Two aspects of electronics engineering that I find more challenging are 1) waiting for circuits to become commercially available and 2) translating performance from simulations to real-world experiments. In many cases, researchers achieve a promising performance when testing circuits and components in a simulated environment but then observe a significant difference in performance when applied in real systems. This can be very frustrating, yet it also motivates us to design new systems that can better cope with reality.

 

Do you collaborate with other universities and researchers worldwide? 

Bartolozzi: I recently coordinated NeuTouch, an EU-funded project that established a doctoral network of 15 students at institutes across Europe, working on different aspects of touch. Students in this network carried out different but complementary studies, ranging from neuroscience studies exploring how the brain processes tactile stimuli to engineering studies aimed at designing circuits for tactile sensors. Some students then applied these concepts to robots and prosthetic devices. One team also conducted a strange experiment, studying how a person could use a third artificial arm and how the signals from this hand are represented in the brain.

I also actively support the neuromorphic community by fostering international collaborations at the Capocaccia workshop, where I serve on its scientific board. During this 15-day workshop, Ph.D. students, postdocs, and faculty from various institutes discuss important topics in neuroscience and brain-inspired engineering.

 

In light of Women’s History Month, what do you wish other young women interested in this field knew? 

Bartolozzi: I am currently the committee chair of one of the IEEE’s circuits and system societies called Women in Circuits and Systems. We are closely considering what we can do to help women and minorities who are underrepresented within the electronic designer community.

It’s not as easy for women or underrepresented communities to become part of a network. So, I would stress the importance of having a network of peers and faculty members who know your work because they see you at conferences and meetings. While this can be challenging especially for reserved individuals, it can help to exchange ideas, get support, and initiate collaborations. Supervisors can support students with this, and if they don’t, it would be wise to ask them for help.

 

What has your experience looked like as a woman in electrical engineering? 

Bartolozzi: I sometimes felt that it was harder for me to be heard during meetings, and that some things said by male colleagues were taken into greater consideration. This could be because I was the only woman in the room or because I wasn’t pushing as hard to be heard. I was always one of the few women around and it felt important for me to be there, to show new generations that women could get there, too.

I also observed differences in salary and some unsavory comments about how women attained professional positions. Thankfully this was from a few individuals, and I don’t think it’s fair to generalize. I was also lucky because my Ph.D. supervisor was always very supportive and helped me to enter the community, encouraging me during conferences and introducing me to his peers. I now have a very strong network because of this.

 

Finally, what research feats do you take the most pride in? 

Bartolozzi: One of the things I take most pride in is the design of the synapse introduced in my Ph.D. thesis that is still being used after many years. Another thing I’m proud of are the achievements of one of the Ph.D. students I supervised, who recently received a personal grant from the European Commission, funding her post-doc work. I supported her in writing the grant, offering feedback, and connecting her with others within our community. This feels like a big professional achievement for me, and I am very proud of her. Finally, I am very proud of the work I have been supervising recently, focusing on low latency and event-driven vision for robots.

 

Bartolozzi is coordinating various interesting research projects at IIT aimed at developing next-generation brain-inspired chips and sensors. She is an esteemed member of the Institute of Electrical and Electronics Engineers (IEEE) and has coordinated various European and international projects, including the European Training Network “NeuTouch” that explores different aspects of touch. Doctoral research partners for this project hailed from Bielefeld in Germany, Sheffield in the U.K., Gothenburg in Sweden, EPFL in Switzerland, Pal Robotics in Spain, and SISSA in Trieste (Northern Italy). Bartolozzi also coordinated Capocaccia, a Sardinia-based neuromorphic computing workshop, and participated in an NSF-funded neuromorphic cognition engineering workshop in Telluride. 

You can visit her IIT profile to stay up to date with her research and learn more about her work.

 

 

All images used courtesy of IIT.



Source link

Leave a comment