Andrea Bajcsy

Timeline

1. 2015–2016

   Robotics at Maryland and Berkeley

   In 2015, she studies robotic perception with Yiannis Aloimonos, director of the Computer Vision Laboratory at the University of Maryland. She spends summer 2015 researching with Ronald Fearing in UC Berkeley's Biomimetic Millisystems Lab. She earns her BS in computer science at the University of Maryland, College Park, in 2016.

2. 2017

   Learning from human corrections

   In 2017, she coauthors Learning Robot Objectives from Physical Human Interaction with Dylan Losey, Marcia O’Malley, and Anca Dragan. The work develops a method for robots to learn task objectives from physical human interventions and tests it in a user study.

3. 2021–2022

   NVIDIA and a Berkeley doctorate

   In 2021, she interns in NVIDIA's Autonomous Driving Research group, led by Marco Pavone. She completes her PhD in electrical engineering and computer sciences at UC Berkeley in August 2022, advised by Claire Tomlin and Anca Dragan, and starts a Berkeley postdoctoral position in August 2022. Her dissertation, Bridging Safety and Learning in Human-Robot Interaction, develops a Bayesian monitor that assesses whether a robot's learned model of human behavior can explain observed human data.

4. September 2023

   Joining Carnegie Mellon

   She joins Carnegie Mellon's Robotics Institute as an assistant professor in September 2023.

5. October 2023

   Learning visual robot rewards

   In October 2023, she coauthors What Matters to You? Towards Visual Representation Alignment for Robot Learning with Ran Tian, Chenfeng Xu, Masayoshi Tomizuka, and Jitendra Malik. The paper introduces Representation-Aligned Preference-based Learning (RAPL), which uses human feedback to align robots' visual representations and learn rewards that reflect user preferences.

6. 2025

   Latent Safety Filters

   In 2025, she receives an NSF CAREER Award for Formalizing Open World Safety for Interactive Robots. With her doctoral student Kensuke Nakamura, she introduces Latent Safety Filters, which use visual inputs and a generative world model to anticipate and prevent unsafe outcomes. She also receives the DARPA Young Faculty Award and Amazon Research Award in 2025.

7. 2026

   Packard and AI2050 fellowships

   In 2026, she receives a Packard Fellowship for Science and Engineering and becomes an AI2050 Early Career Fellow. She is an assistant professor in Carnegie Mellon's Robotics Institute and leads the Interactive and Trustworthy Robotics Lab (Intent Lab), where her research develops safe robot interaction policies and aligns AI models with users' values.