Yiwei Lyu
Carnegie Mellon University
yiweilyu@andrew.cmu.edu
Bio
Yiwei Lyu is a final-year PhD student in the Department of Electrical and Computer Engineering at Carnegie Mellon University, where she works with Dr. John Dolan. Her research interests in robotics include safety-critical control, behavior planning, and integration of human factors in robot decision-making processes. Her research aims to develop robot behaviors that are both safe and efficient, supported by provable guarantees. Yiwei is the recipient of the Qualcomm Innovation Fellowship and was recognized as a Human-Robot Interaction (HRI) Pioneer in 2024. Her research has received best paper awards and nominations at various conferences, including AAMAS, AAAI, IFAC CPHS, as well as workshops at ICRA and IJCAI. Prior to joining CMU in 2019, Yiwei earned her Bachelor’s degree in Electronic and Information Engineering from the Chinese University of Hong Kong, Shenzhen (CUHK-Shenzhen).
Areas of Research
- Robotics
Enabling Provable Safety for Robot Autonomy
As robots become increasingly integrated into our daily lives, ensuring their safe and autonomous operation is of paramount importance. My research aims to develop decision-making algorithms that enable autonomous systems to proactively identify and mitigate potential risks, with safety guarantees backed by theoretical foundations and empirical validation. Specifically, my work explores innovative integrations of control theory, machine learning, and social science to tackle three principal challenges in robot safety: 1) uncertainty in rapidly changing environments, 2) individual differences in companion agents during interactions, and 3) misaligned safety expectations when robots transition from being mere tools to active participants in human society. My research designs methodologies that address not only explicit pre-computed safety specifications but also implicit safety notions learned and reasoned from human demonstrations. These systems are tailored to adapt seamlessly to the variability in behaviors and capabilities of human and robotic agents, thereby enhancing collective safety and task performance during interactions. By establishing provable safety for robot autonomy, I aim to foster trust and acceptance of robotic applications, such as self-driving cars in mixed-autonomy settings and collaborative human-robot manufacturing, while also providing foundational insights for future legal and ethical studies on standards and regulations for robot safety that extend beyond physical configurations.