The Engineer's Impact - Soodeh Nikan

Your inside look at faculty’s research and its effect on society

In this new Q&A series, we’ll feature Western Engineering faculty members to gain a succinct overview of their research, understand its impact on society, and discover intriguing little-known facts.

Meet Electrical and Computer Engineering Assistant Professor Soodeh Nikan. 

Zhao-profile-pic.pngCan you describe your research?

My research focuses on developing explainable artificial intelligence (AI) strategies for autonomous vehicles, with learning abilities that imitate human brain perception and reasoning. The goal of my research is to design robust, unbiased and industry-standardized AI-based platforms for intelligent and safe mobility which can be extended to other vision-based applications.

How does your research impact society in everyday life?

Autonomous driving not only increases road safety with more accuracy and less latency than humans but is also beneficial to society by improving the comfort level of mobility for the elderly and people with disability. As the market moves from “Partial Autonomy” to “Full Automation”, more control will shift from the human driver to the vehicle. Requirements at higher levels are not only about occupant safety but also convenience, comfort, and high-quality user experience. In my research program, the goal is to equip cars with an intelligent vision-based system using AI technology for safe and smooth vehicle control which aids in reducing fatal road collisions. My research is focused on designing compact and optimized deep neural network (DNN) architectures not only to make them more sustainable with reduced computational cost and carbon footprint but also capable of supporting latency constraints of edge computing devices. An extremely exciting avenue of my research is to cope with uncertainty, robustness, and reasoning issues of DNN models which is critical in safety-related applications in industries such as automotive and medical.  

What’s an interesting, little-known fact related to your research?

Thanks to the evolution of AI technologies in the automotive sector, the dream of self-driving cars has become very close to a commercial reality which is going to be a life-changing product. However, converting deep neural networks from a magic black box to a transparent post hoc - which will increase reliability and certainty - are critical factors in safety-related domains and this has not been completely solved yet.