About

Eric Frew is a Professor in the Department of Aerospace Engineering Sciences at the University of Colorado Boulder, where he has been serving since 2018. His research focuses on networked heterogeneous unmanned aircraft systems, optimal distributed sensing by mobile robots, controlled mobility in ad-hoc sensor networks, miniature self-deploying systems, and guidance and control of unmanned aircraft in complex atmospheric phenomena. He previously directed the Research and Engineering Center for Unmanned Vehicles (RECUV) from 2012 to 2018 and the Autonomous Systems Interdisciplinary Research Theme (ASIRT) from 2018 to 2022. Frew earned his PhD in Aeronautics and Astronautics from Stanford University in 2003, following a Master’s degree in the same field from Stanford in 1996, and a Bachelor’s degree in Mechanical Engineering from Cornell University in 1995. His professional experience includes postdoctoral research at the University of California, Berkeley, and faculty roles at the University of Colorado, where he has also served as Associate Chair for Departmental Affairs. His contributions to the field have been recognized through numerous awards, including the NSF CAREER Award, the Outstanding Mentor Award for Faculty Mentoring at the University of Colorado, and the H. Joseph Smead Faculty Fellow. Frew’s work is characterized by a focus on advancing autonomous systems and unmanned vehicle technologies.

Research topics

• Computer Science
• Artificial Intelligence
• Engineering
• Distributed computing
• Aerospace engineering
• Human–computer interaction
• Meteorology
• Systems engineering
• Environmental science
• Computer Security
• Geography
• Aeronautics
• Ecology
• Telecommunications
• Data science
• Simulation
• Computer network
• Multimedia
• Physics

Selected publications

• TORUS-LItE: RAAVEN UAS Data. Version 1.0

  Open MIND · 2026-01-01

  datasetOpen access

  Flight level meteorological and aircraft state data from the University of Colorado RAAVEN (Robust Autonomous Aerial Vehicle - Endurant and Nimble) sUAS (small Unmanned Aircraft System) platform that flew during the TORUS-LItE (Targeted Observations using Radars and UAS in Supercells Left Flank Intensive Experiment ) campaign over the plains region of the United States. The UAS flew near inflow and left flank missions in and around supercell thunderstorms.

  PublisherOA PDFDOI

• Competency self-assessment for a learning-based autonomous aircraft system

  Frontiers in Aerospace Engineering · 2025-02-14

  articleOpen access

  Introduction Future concepts for airborne autonomy point toward human operators moving out of the cockpit and into supervisory roles. Urban air mobility, airborne package delivery, and military intelligence, surveillance, and reconnaissance (ISR) are all actively exploring such concepts or currently undergoing this transition. Supervisors of these systems will be faced with many challenges, including platforms that operate outside of visual range and the need to decipher complex sensor or telemetry data in order to make informed and safe decisions with respect to the platforms and their mission. A central challenge to this new paradigm of non-co-located mission supervision is developing systems which have explainable and trustworthy autonomy and internal decision-making processes. Methods Competency self-assessments are methods that use introspection to quantify and communicate important information pertaining to autonomous system capabilities and limitations to human supervisors. We first discuss a computational framework for competency self-assessment: factorized machine self-confidence (FaMSeC). Within this framework, we then define the generalized outcome assessment (GOA) factor, which quantifies an autonomous system’s ability to meet or exceed user-specified mission outcomes. As a relevant example, we develop a competency-aware learning-based autonomous uncrewed aircraft system (UAS) and evaluate it within a multi-target ISR mission. Results We present an analysis of the computational cost and performance of GOA-based competency reporting. Our results show that our competency self-assessment method can capture changes in the ability of the UAS to achieve mission critical outcomes, and we discuss how this information can be easily communicated to human partners to inform decision-making. Discussion We argue that competency self-assessment can enable AI/ML transparency and provide assurances that calibrate human operators with their autonomous teammate’s ability to meet mission goals. This in turn can lead to informed decision-making, appropriate trust in autonomy, and overall improvements to mission performance.

  PublisherOA PDFDOI

• Formal Verification of Unknown Dynamical Systems via Gaussian Process Regression

  IEEE Transactions on Automatic Control · 2025-01-22 · 2 citations

  article

  Leveraging autonomous systems in safety-critical scenarios requires verifying their behaviors in the presence of uncertainties and black-box components that influence the system dynamics. In this work, we develop a framework for verifying discrete-time dynamical systems with unmodeled dynamics and noisy measurements against temporal logic specifications from an input–output dataset. The verification framework employs Gaussian process (GP) regression to learn the unknown dynamics from the dataset and abstracts the continuous-space system as a finite-state, uncertain Markov decision process (MDP). This abstraction relies on space discretization and transition probability intervals that capture the uncertainty due to the error in GP regression by using reproducible kernel Hilbert space analysis as well as the uncertainty induced by discretization. The framework utilizes existing model checking tools for verification of the uncertain MDP abstraction against a given temporal logic specification. We establish the correctness of extending the verification results on the abstraction created from noisy measurements to the underlying system. We show that the computational complexity of the framework is polynomial in the size of the dataset and discrete abstraction. The complexity analysis illustrates a tradeoff between the quality of the verification results and the computational burden to handle larger datasets and finer abstractions. Finally, we demonstrate the efficacy of our learning and verification framework on several case studies with linear, nonlinear, and switched dynamical systems.

  PublisherDOI

• Wind Aware Batch Informed Trees for Path Planning of Small UAS to Minimize Travel Time

  2025-07-16

  article

  Battery capacity limits the endurance of small unmanned aerial systems (UAS). This paper presents a wind-aware extension of the Batch Informed Tree (BIT*) planner that exploits both ambient horizontal winds and thermal updrafts to reduce travel time. A wind-informed heuristic guides heading and airspeed adjustments while ensuring collision-free paths in cluttered environments. Experiments were conducted in both a 2D urban environment and a 3D cluttered environment with thermal updrafts to validate the approach. In simulations over these scenarios, travel-time reductions range from about 3.5 % under mild winds to as much as 25 % when strong updrafts are available, for only a 1.5 % increase in path length. These results illustrate how even modest wind fields can be leveraged to extend mission duration over kilometer-scale flights. A notable drawback is the increased computation time. A planning runs up to 70–80 times slower than a simple distance-minimizing strategy. Future work will aim to accelerate the planner and relax the requirement for perfect wind-field knowledge, moving toward real-time applicability.

  PublisherDOI

• 3D Path-Following Guidance via Nonlinear Model Predictive Control for Fixed-Wing Small UAS

  ArXiv.org · 2025-12-30

  articleOpen accessSenior author

  This paper presents the design, implementation, and flight test results of two novel 3D path-following guidance algorithms based on nonlinear model predictive control (MPC), with specific application to fixed-wing small uncrewed aircraft systems. To enable MPC, control-augmented modelling and system identification of the RAAVEN small uncrewed aircraft is presented. Two formulations of MPC are then showcased. The first schedules a static reference path rate over the MPC horizon, incentivizing a constant inertial speed. The second, with inspiration from model predictive contouring control, dynamically optimizes for the reference path rate over the controller horizon as the system operates. This allows for a weighted tradeoff between path progression and distance from path, two competing objectives in path-following guidance. Both controllers are formulated to operate over general smooth 3D arc-length parameterized curves. The MPC guidance algorithms are flown over several high-curvature test paths, with comparison to a baseline lookahead guidance law. The results showcase the real-world feasibility and superior performance of nonlinear MPC for 3D path-following guidance at ground speeds up to 36 meters per second.

  PublisherOA PDF

• Targeted Observation by Radars and UAS of Supercells: TORUS

  Bulletin of the American Meteorological Society · 2025-10-31

  articleSenior author

  Abstract Targeted Observation by Radars and Uncrewed Aircraft Aystems (UAS) of Supercells (TORUS) aimed to improve the conceptual model of supercell thunderstorms through advancing the understanding of the role of storm-generated airmass boundaries and coherent structures in the development of near-surface rotation. Research questions guiding the field phase of TORUS focused on left-flank vertical vorticity sheets, streamwise vorticity currents, left-flank convergence boundaries, and rear-flank internal surges. Research questions also aimed to address the relationship between inflow modification and supercell characteristics. Across three field seasons (2019, 2022, and 2023), data on 46 supercell thunderstorms were collected through coordinated deployments of radars, lidars, mobile mesonets, UAS, manned aircraft, radiosondes, and swarmsondes. More than 200 scientists and engineers (many of whom were students) participated in the TORUS field deployments. The scientific motivation for TORUS, experiment design, and examples of data/analysis are presented in this article. Significance Statement Targeted Observation by Radars and Uncrewed Aircraft Systems (UAS) of Supercells (TORUS) was a collaborative research project funded by the National Science Foundation and the National Oceanic and Atmospheric Administration to advance understanding of supercells. TORUS involved more than 200 scientists and engineers (many of whom were students) who led data collection on 46 supercell thunderstorms across three field seasons. This effort constituted the most deployments of UAS within supercells and, on 17 May 2019, likely yielded the longest continuous airborne multi-Doppler radar sampling of a Great Plains supercell ever conducted.

  PublisherDOI

• “A Good Bot Always Knows Its Limitations”: Assessing Autonomous System Decision-Making Competencies through Factorized Machine Self-Confidence

  ACM Transactions on Human-Robot Interaction · 2025-04-28 · 4 citations

  article

  How can intelligent machines assess their competency to complete a task? This question has come into focus for autonomous systems that algorithmically make decisions under uncertainty. We argue that machine self-confidence—a form of meta-reasoning based on self-assessments of system knowledge about the state of the world, itself, and ability to reason about and execute tasks—leads to many computable and useful competency indicators for such agents. This article presents our body of work, so far, on this concept in the form of the Factorized Machine Self-Confidence (FaMSeC) framework, which holistically considers several major factors driving competency in algorithmic decision-making: outcome assessment, solver quality, model quality, alignment quality, and past experience. In FaMSeC, self-confidence indicators are derived via “problem-solving statistics” embedded in Markov Decision Process solvers and related approaches. These statistics come from evaluating probabilistic exceedance margins in relation to certain outcomes and associated competency standards specified by an evaluator. Once designed, and evaluated, the statistics can be easily incorporated into autonomous agents and serve as indicators of competency. We include detailed descriptions and examples for Markov Decision Process agents and show how outcome assessment and solver quality factors can be found for a range of tasking contexts through novel use of meta-utility functions, behavior simulations, and surrogate prediction models. Numerical evaluations are performed to demonstrate that FaMSeC indicators perform as desired (references to human subject studies beyond the scope of this article are provided).

  PublisherDOI

• Autonomous Planning for Targeted Observation of Severe Weather

  Lecture notes in computer science · 2025-08-25

  book-chapterSenior author
  PublisherDOI

• 3D Path-Following Guidance via Nonlinear Model Predictive Control for Fixed-Wing Small UAS

  arXiv (Cornell University) · 2025-12-30

  preprintOpen accessSenior author

  This paper presents the design, implementation, and flight test results of two novel 3D path-following guidance algorithms based on nonlinear model predictive control (MPC), with specific application to fixed-wing small uncrewed aircraft systems. To enable MPC, control-augmented modelling and system identification of the RAAVEN small uncrewed aircraft is presented. Two formulations of MPC are then showcased. The first schedules a static reference path rate over the MPC horizon, incentivizing a constant inertial speed. The second, with inspiration from model predictive contouring control, dynamically optimizes for the reference path rate over the controller horizon as the system operates. This allows for a weighted tradeoff between path progression and distance from path, two competing objectives in path-following guidance. Both controllers are formulated to operate over general smooth 3D arc-length parameterized curves. The MPC guidance algorithms are flown over several high-curvature test paths, with comparison to a baseline lookahead guidance law. The results showcase the real-world feasibility and superior performance of nonlinear MPC for 3D path-following guidance at ground speeds up to 36 meters per second.

  PublisherDOI

• Autonomous Uncrewed Aircraft for Mobile Operations in Severe Weather

  Lecture notes in computer science · 2025-08-25

  book-chapter
  PublisherDOI

Recent grants

• I/UCRC Phase I: Center for Unmanned Aircraft Systems

  NSF · $686k · 2012–2018

• Collaborative Research: RAPID--Integration of Unmanned Aircraft System (UAS) into the Program for Research on Elevated Convection with Intense Precipitation

  NSF · $55k · 2015–2017

• IUCRC: Phase 2: Center for Unmanned Aircraft Systems

  NSF · $919k · 2017–2022

• RI: Small: Providing Quality of Information in Robot Sensor Networks

  NSF · $500k · 2011–2015

• I/UCRC FRP: Collaborative Research: Network-enabled Airborne Autonomy

  NSF · $120k · 2015–2016

Frequent coauthors

• Brian Argrow

  University of Colorado Boulder

  63 shared

• Jack Elston

  University of Illinois Urbana-Champaign

  33 shared

• Cory Dixon

  University of Colorado Boulder

  32 shared

• Maciej Stachura

  Black Swift Technologies (United States)

  27 shared

• Katherine Glasheen

  University of Colorado Boulder

  26 shared

• J.D. Jackson

  24 shared

• Nisar Ahmed

  22 shared

• Morteza Lahijanian

  University of Colorado Boulder

  22 shared

Education

• B.S.

  Cornell University

  1995

• M.S., Aeronautics and Astronautics

  Stanford University

  1996

• Ph.D., Aeronautics and Astronautics

  Stanford University

  2003

Awards & honors

• Outstanding Mentor Award for Faculty Mentoring, University o…
• H. Joseph Smead Faculty Fellow, 2014
• American Institute of Aeronautics and Astronautics (AIAA) As…
• Provost's Faculty Achievement Award, 2009
• NSF Faculty Early Career Development (CAREER) Award, 2009