CSL AI research to transform everyday life

In the 1970s, then CSL Director Robert Chien led an interdisciplinary research team that conducted some of the first work on artificial intelligence (AI) on the University of Illinois Urbana-Champaign (UIUC) campus. Since those early contributions, CSL has continued to lead the way in the field of AI.

Today, the interdisciplinary environment CSL provides makes it the perfect place for researchers from a wide variety of UIUC units to conduct AI research.

“Many of the novel AI algorithms and innovative AI frameworks, systems, and applications developed by CSL researchers follow the CSL tradition of excellence,” said Klara Nahrstedt, CSL director. “The strong intersection between AI theory and AI system research, collaboration of AI researchers from across campus, and the constant curiosity and dedication of our researchers to solve hard AI problems are all factors leading to the high-impact AI innovations coming out of CSL.”

The uses and applications of AI have reached into nearly every aspect of modern-day life. From agricultural equipment to medical treatments to electronic devices and much more, CSL researchers are pushing the boundaries of AI research.

 Agriculture
In 2020, the National Science Foundation and the US Department of Agriculture announced seven new AI institutes across the country. Two of them are located in UIUC’s Grainger College of Engineering, and one is led by CSL faculty member Vikram Adve. The AI Institute for Future Agricultural Resilience, Management, and Sustainability (AIFARMS) has the goal of advancing research in computer vision, machine learning, soft object manipulation, and intuitive human-robot interaction. The institute is intended to solve major agricultural challenges including labor shortages, efficiency and welfare in animal agriculture, environmental resilience of crops, and the need to safeguard soil health.

One location where AIFARMS research will be carried out is the Illinois Autonomous Farm led by Girish, led by CSL’s Girish Chowdhary, professor of agricultural and biological engineering. The working farm will give AI researchers a place to evaluate their methods and build up the infrastructure needed for their research.

“The overall goal of our work is to improve sustainability in agriculture by replacing chemicals with mechanical alternatives where we can, reducing chemicals where we can, and increasing diversity and productivity of Illinois agriculture by using robots,” said Chowdhary.

Personalized medicine
Currently, patients are often prescribed medical treatments based on what has worked for the majority of other patients with the same ailment. That approach is less than ideal, because each individual will respond to medications and therapies differently. The use of AI to diagnose and treat a variety of diseases and ailments through individualized medicine would be a paradigm shift for the entire healthcare sector, and would greatly improve the success and efficacy of treatments. CSL faculty member Ravi Iyer and his students have investigated the use of AI and machine learning to develop individualized medicine for a number of maladies, including epilepsy, depression, and even cancer.

Recent reductions in the cost of analyzing the human genome have made individualized medicine a more realistic possibility. Genetics and other factors, such as patient history, can be used by AI tools to generate predictions of treatment efficacy that serve as actionable intelligence, helping medical providers select the best treatment option for each individual patient.

“Innovations in artificial intelligence methodologies capable of harnessing the power of clinical and genomic data have shown early success in fulfilling pharmacogenomic promises of individualized medicine,” said Iyer, professor of electrical and computer engineering (ECE). “Methods embodying artificial intelligence play a crucial role in the generation of actionable intelligence and personalized medicine.”

COVID-19
When the pandemic first broke out in the spring of 2020, a number of CSL faculty jumped into action to come up with AI solutions to combat COVID-19. The C3.ai Digital Transformation Institute (DTI), co-led by CSL’s R. Srikant, funded 26 projects on the topic, five of which were led by CSL researchers.

“I am delighted to see the interdisciplinary nature of the projects led by the CSL faculty members,” said Srikant, professor of ECE. “The collaboration between engineers, computer scientists, epidemiologists, and medical/healthcare professionals will produce both immediate and lasting benefits to society.”

The projects being led by CSL faculty are focused on helping hospital systems, improving patient treatment, controlling the virus, and keeping patient data safe. The specific topics are:

• Developing a best practices guidance system to address hospital staff shortages.
• Using AI and machine learning on audio-visual data to predict whether a patient’s condition will deteriorate or improve, saving hospital space for those who need it most.
• Developing algorithms to combine real-time testing data with epidemiological models to better inform decision-makers on what effect control strategies have on the spread of the coronavirus.
• Using AI, machine learning, and cryptography to protect private patient medical data.
• Studying the virus’s genomic information to determine what separates SARS-CoV-2 from other viruses that haven’t caused global pandemics, to better prepare for future outbreaks.

Srikant is joined on the C3.ai DTI leadership team by CSL’s Tandi Warnow, who is a professor of computer science and the Institute’s co-chief scientist.

Electronic devices
Enhancing so-called “edge” devices, which can include cell phones, smart watches, and other IoT devices, with AI capabilities is a major goal for researchers in both industry and academia. These devices generate huge volumes of sensory data from their built-in sensors in the form of cameras, microphones, gyroscopes, and other technology. Processing that data is challenging due to the limited computational resources and constrained energy supplies of edge devices. A team from CSL led by Naresh Shanbhag is working to improve the energy efficiency and functionality of these devices.

“These data-generating devices can be made a lot smarter if they could be AI enabled,” said Shanbhag, professor of ECE. “The mainstream method of using AI today is via deep neural networks (DNNs). Bringing DNNs into devices such as cell phones is extremely challenging due to the need to operate on a tight energy budget with severely limited storage and computational capacity.”

Unfortunately, when energy efficiency and accuracy are increased in these devices, their networks are more easily compromised. Finding out why this happens is one of the main goals of Shanbhag’s research team.

Autonomous driving
One of the most famous applications of AI technology is in autonomous vehicles. In early 2020, CSL introduced the High Bay, a new space for research in this area. The High Bay is located in the University of Illinois Urbana-Champaign Research Park and includes space for autonomous vehicles, along with access to areas that allow researchers to drive in a road-like environments.

These additions were crucial to the expansion of the Center for Autonomy (CfA), a research group run out of CSL.

“Previously, we just didn’t have space for many of these activities — they just didn’t happen on our campus — and creating the space has provided significant support for our faculty to develop an autonomous driving program,” said CSL’s Geir Dullerud, CfA director and professor of mechanical science and engineering.

In addition to the space, CfA added a second vehicle to its autonomous fleet, an off-road vehicle donated by John Deere. Both vehicles in the fleet are being used by a number of classes taught by CSL/CfA faculty.