Kate Gleason College of Engineering

Dr. Azadeh-Fard uses quantitative analytical methods and machine learning techniques on large healthcare data sets to understand the relationship between patient/hospital characteristics and their ultimate healthcare outcomes. Her work provides insights into the factors that influence the length of stay, in-hospital mortality, and readmission of patients with COPD and congestive heart failure. These insights will enable healthcare systems to better analyze and mitigate risks to improve patient outcomes.

Dr. Baheri focuses on research at the intersection of autonomy, controls, and machine learning to improve the safety of autonomous systems. His group explores how reinforcement learning – a machine learning technique that rewards desired behaviors – can be deployed, verified, and validated. His work seeks to advance the development of safe, certified, and efficient intelligent autonomous systems like self-driving cars, unmanned aerial vehicles, and collaborative robots.

Dr. Díaz A. uses predictive algorithms to investigate the self-assembly of colloids. His team uses trained neural networks to detect and categorize colloids of various shapes in microscopic images. The images and tracked datasets are then used in combination with physical models to explore the dynamics of fundamental phenomena like packing and crystallization. A better understanding of these fundamental phenomena will ultimately enable the development of new nanostructured materials with unique properties.

Dr. Heard is using state-of-the-art AI techniques (i.e., multi-modal data fusion, model compression, reinforcement learning, generative modeling) to promote fluent team collaborations between humans and robots in high-stress, dynamic domains, such as search-and-rescue. In this work, machine-learning paradigms enable intelligent robotic adaptations based on robust estimates of the human’s internal state to optimize team performance. This research may lead to human-specific emergent robot behaviors that promote natural, fluid team interactions that allow robots to be used in a variety of environments.

Dr. Kolodziej utilizes machine learning (ML) to assess the health of engineering components and systems, such as flight actuators, gas compressor valves, seals, and ventricular heart pumps. In his work, Dr. Kolodziej measures the sound or vibrations emanating from the system during operation and then analyzes the collected data sets using signal processing and machine learning techniques to diagnose the health of the equipment. Broadly applicable to many different components and systems, this work is of great interest to industry because of its potential to reduce the costs associated with downtime and improve safety by detecting faults in safety-critical components.

Dr. Kuhl, in collaboration with Amlan Ganguly, Andres Kwasinski, and Clark Hochgraf, is using AI in the design of intelligent Material Handling Systems (iMHS) including task selection, path planning, localization, navigation, and communication of autonomous mobile robots in a warehouse environment. AI applications in iMHS could lead to increased productivity, efficiency, safety, reliability, and robustness of material handling and supply chain systems.

Dr. Kwasinski is using AI to realize wireless devices that are self-aware of their operating situation and can learn how to adapt their operation to the conditions in a wireless network. Analogous to self-driving vehicles that learn how to move around in our physical 3D world, Dr. Kwasinski’s research into cognitive radios enables wireless devices that learn by themselves how to autonomously operate in the environment of the radio spectrum. With their capacity to develop awareness of the operating conditions and to adapt their operation to the environment, cognitive radios can make wireless communications more efficient and resilient, providing ubiquitous radio communications that are tuned to the end-user needs.

Dr. Dongfang Liu’s research uses artificial intelligence and deep learning to address real-world challenges with societal relevance. Examples include autonomous driving, mobile agents, smart transportation, and geospatial computing. Ongoing projects include autonomous driving perception and video instance segmentation. The algorithmic and analytical products developed by Dr. Liu contribute to the advancement of contemporary autonomous robots and computer vision research.

Dr. Liu is working on developing a flexible, cost effective, and accurate machining monitoring system to detect cutting tool conditions and relevant machining situations using AI techniques. In this work, AI is used as a decision-making algorithm to detect machining problems from various monitored signals. This research work could lead to a smart monitoring solution to meet the requirements of Industry 4.0 in the machining industry.

Dr. Lyshevski is analyzing information governance, complexity, and data processing in physical and cyber domains to empower system intelligence, data quality, and cybersecurity. Parametric and non-parametric statistical models, heterogeneous Markov-Turing descriptive paradigm, and adaptive control schemes are examined, applied, and validated. The work applies new approaches, algorithms, design schemes, and technology developments to legacy systems and new autonomous intelligent platforms. This work will enable dynamic data-driven intelligent decision-and-controls and management of cyber-physical defense systems.

Dr. McConky is using AI techniques to solve large combinatorial optimization problems. In this work, deep learning techniques are used to solve problems with exponential decision spaces in a fast and efficient manner. As one example, we can train a single model and quickly provide quality solutions to team-orienteering problems of various sizes. This work has applications in improving the speed and quality of logistics solutions, such as the scheduling of transportation routes or optimizing kidney exchange networks.

Dr. Merkel is collaborating with the National Technical Institute for the Deaf to improve the performance of smart assistants for Deaf and Hard-of-Hearing (DHH) users. Current smart assistant devices such as Amazon Alexa and Apple Siri are biased toward users with “typical” speech patterns. In this work, researchers are gathering a large dataset of speech samples from DHH individuals with low speech intelligibility. The data will be used to develop new techniques that can augment the behavior of state-of-the-art automatic speech recognition algorithms for atypical speech.

Dr. Padmanabhan is using AI techniques to accelerate the development of new materials that exhibit both solid-like and liquid-like properties, such as toothpaste, yogurt, and bio-injectable gels. In this work, AI techniques are used to design the particle-particle interactions that lead to specific microstructures, which in turn affect the mechanical properties of the materials. As one example, this research could lead to the development of injectable gels that promote wound healing and bone regeneration.

Dr. Proano works at the intersection of data analytics and operations research. He is interested in the use of prescriptive analytics and optimization modeling in healthcare. With a passion for increasing worldwide access to vaccines, he has analytics and optimization models to study and make recommendations for changes to vaccine markets.

Dr. Perez-Raya is developing data and physics-driven artificial intelligence (AI) models for medical imaging, pollution monitoring, and optimization problems. His work focuses on developing techniques that can use sparse and noisy experimental measurements to generate a de-noised high-resolution and reliable solution. As one example, his research could lead to more reliable and effective detection of tumors and diseases from medical images.

Dr. Rashedi uses AI to better understand and monitor human movement. Using data analytics Dr. Rashedi seeks to better predict fall risk among post-stroke patients, measure the efficacy of and improve rehabilitation programs, and study and design improved exoskeleton systems.

Dr. Richards uses convolutional neural networks to develop efficient methods for improving the point-of-care evaluation of pathological conditions associated with mechanical changes of soft tissue, such as cancer, cardiovascular diseases, and musculoskeletal disorders. His team trains neural networks using simulated data, rather than large patient data sets, to quickly develop neural networks with different characteristics for a variety of applications. Deployment of these trained neural networks on measured data, typically acquired from ultrasound imaging systems, can improve the accuracy, efficiency, and usability of these techniques in point-of-care patient diagnosis and monitoring.

Dr. Sahin’s research uses AI and machine learning (ML) techniques for research involving collaborative robotics, novelty detection systems, gesture detection, swarm intelligence, and indoor navigation. In the case of collaborative robotics, his group uses AI and ML techniques to determine the location and mental state (comfort index) of humans working with cobots by measuring location and physiological signals such as heart rate, pupil size, and galvanic skin response. This work will improve the safety and efficiency of collaborative robotic systems. In the case of novelty detection systems, ML and AI are used to determine when machines and living beings are operating outside of their normal limits so that technicians and caretakers can be informed that an intervention is required.

In collaboration with the University of Connecticut, Dr. Yan is using AI techniques to improve the modeling and solving of operation optimization problems in power and manufacturing systems. In this work, AI techniques are used to tighten the mixed-integer linear problem formulations and efficiently solve subproblems in the decomposition and coordination framework. This research could lead to the development of AI-based formulation tightening approaches and AI-based optimization methods to reduce the operation costs of power systems and improve the on-time delivery of manufacturing systems.