Development of an optimized emergency escape maneuver for UAV quad-copters in presence of (human) obstacles
This research involves the development of a method to establish the severity of potential collisions with human obstacles during flight, and establish whether an emergency escape maneuver is necessary (resulting, in case, in a temporary take-over of commands). Once the emergency maneuver is initiated, an optimized reactive collision avoidance scheme is followed until the collision is fully avoided. The reactive controller is developed a-priori using an off-line approximate reinforcement learning method based on a pre-defined reward function tailored to a human geometry and sensitive areas. The implemented algorithm aims to be versatile and adaptable to different UAV dynamics via a set of configurable parameters, taking into account initial dynamic delays and maximum accelerations on each axis. These parameters can be retrieved and configured during flight with the help of a set of calibration maneuvers. Thanks to the estimation of the severity of the collision situation and the use of a reactive avoidance scheme, the algoirthm has been shown to work for multiple obstacles as well as moving obstacles.
Last edited on May 28th, 2015
|Feb 2015 - current
Research Intern at Max Planck Institute for Biological Cybernetics.
Autonomous Robotics and Human-Machine Systems group
Jul 2014 - Feb 2015
R&D Scientist I at Honeywell Aerospace in Brno, Czech Republic
Performed an internship within the Advanced Technology department.
Project 1: Integration of Model Based Engineering within standard development practices for cockpit display systems.
Project 2: Development of multi-modal interaction software for cockpit displays.
|Aug 2013 - Jun 2014
Teaching Assistant at Delft University of Technology
General responsibilities: student support, development of assignments/exams, and grading. Assisted the following courses: Simulation, Verication and Validation (AE3212-II), ComputationalModelling (AE2220-II), Exploring Aerospace Engineering (AE1111-I).
|2013 - Current
M.Sc. in Aerospace Engineering at Delft University of Technology
Honours track student, specializing in Control & Simulation
|Fall Semester 2012
|Minor in Robotics at Nanyang Technological University, Singapore
|2010 - 2013
B.Sc. in Aerospace Engineering at Delft University of Technology
Bachelor Thesis: Design of a controllable system for the guided atmosphere-assisted deceleration of a human-rated precursor vehicle to Mars. Performed within a team at TU Delft under guidance from NASA Langley. Accepted for the 12th International Planetary Probe Workshop in Cologne, Germany, 2015.