Year05/2023 - present

Year05/2021 - 04/2023

Year03/2017 - 04/2021, 50% since 10/2018, 75% from 10/2019 to 05/2020

SummaryFully autonomous drones. See Technology and some independent in-depth reviews.

• Model-based high-speed and agile predictive trajectory tracking control for the first fully autonomous consumer flying robot with onboard surround obstacle avoidance and advanced motion planning and person tracking (Skydio R1).
• Aerodynamics modeling, system identification, fault detection. Online wind estimation, collision detection and reflexes deployed in a commercial product (Skydio R1, Skydio 2, Skydio X2).
• Development of a beyond-state-of-the-art model-predictive flight planning and control control system incorporating wind, modeling errors, actuation limits and obstacles (Skydio 2, Skydio X2).
• Development of Field-Oriented Control (FOC) electronic speed control (ESC) for brushless DC (BLDC) motors on an STM32F0 based embedded platform for drones, deployed in commercial products (Skydio 2, Skydio X2).
• Controller implementation on embedded computational platforms in C++. Data analysis from logs, model exploration, and machine learning using Python.

Year10/2018 - 05/2020 (50% from 10/2018 to 09/2019, 25% from 10/2019 to 05/2020)

SummarySee Publications.

SummarySee Master's Thesis in Publications.

2019Georges Giralt PhD award for best robotics dissertation in Europe

2018IROS Conference Best Paper Award Winner, Robust Visual-Inertial State Estimation with Multiple Odometries and Efficient Mapping on an MAV with Ultra-Wide FOV Stereo Vision

2013IROS RoboCup Best Paper Award Winner, Stereo vision based indoor/outdoor navigation for flying robots

2020RSS 2020 Workshop on Perception and Control for Fast and Agile Super-Vehicles [video], virtual

2019Universitz of Zurich RPG Lab | ETH Zurich RSL | Learning to fly meetup: Fully Autonomous Vision Based Flight in the Wild: Progress and Challenges, Zurich, Switzerland

2019RSS 2019 Aerial Interaction and Manipulation: Unsolved Challenges and Perspectives: Towards Learning Manipulation Skills Under Wind Influence, Freiburg, Germany

2018IEEE CDC 2018 special session, Miami Beach, FL, USA

2017Bay Area Robotics Symposium (BARS), Berkeley, CA, USA

2016RSS Workshop on Human-Robot Interaction for Small and Personal Unmanned Aerial Vehicles, Ann Arbor, MI, USA

2014RSS Workshop on Resource-efficient Integration of Planning and Perception for True Autonomous Operation of Micro Air Vehicles (MAVs), Berkeley, CA, USA

A. EditorAssociate Editor (AE) for IEEE International Conference on Robotics and Automation (ICRA) 2020, 2021, 2022, 2023

ReviewerIEEE International Conference on Robotics and Automation (ICRA)

ReviewerIEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)

ReviewerIEEE Transactions on Robotics (T-RO)

ReviewerIEEE Robotics and Automation Letters (RA-L)

ReviewerThe International Journal of Robotics Research (IJRR)

ReviewerEuropean Control Conference (ECC)

ReviewerAmerican Control Conference (ACC)

ReviewerIEEE/ASME International Conference on Advanced Intelligent Mechatronics (AIM)

ReviewerJournal of Aerospace Engineering (JAERO)

AwardsGeorges Giralt PhD Award 2019 for best European doctoral thesis in robotics

ThesisModel-Based Control of Flying Robots for Robust Interaction under Wind Influence [ pdf ]. Control methods to make autonomous MAVs flying in unknown and uncertain environments robust to environmental forces. Covers interaction control; collision detection, identification, isolation and reflexes; tactile mapping; aerodynamics modeling; wind speed estimation and compensation. See Publications.

AwardsGraduated with Honors (top 5% in class)

ThesisRobust control of rigid body motion [ PDF ] First-order sliding mode control with boundary layer and PD sliding surface combined with distubance observer control applied to the model of an unmanned underwater vehicle (UUV) under strong wave disturbances. Developed an iterative control allocation algorithm for vectored propulsion with saturation. Dynamics of UUV derived using tensor representation. Feedback control using Euler-Rodriguez (dual quaternion) representation of body configuration. Developed while working on the Hipersfera airship.