UAVs, Navigation and Sensor Orientation

The navigation is composed of two main concepts: positioning & guidance.

  • The positioning is the determination of the position and velocity of a moving object with respect to a known reference
  • The guidance is the planning and maintenance of a course from one location (origin) to another (destination)

TOPO is active in aerial photogrammetry research, both airborne and drone based. In this context, many challenging problems arise related to sensor orientation determination, autonomous navigation, sensor fusion, design, testing and operation of micro aerial vehicles, both fixed wing and multi-rotor.

A list of currently available semester/master projects is maintained below.

Collaborative Navigation in UAVs

collaborative mapping
A Simulation of two UAVs performing collaborative mapping of a natural canyon.

Global Navigation Satellites Systems (GNSS), such as GPS, are currently the mainstream way to obtain position control in unmanned aerial vehicles. However, GNSSs become less reliable or unavailable in cluttered environments where the satellites may not be visible or where multipath might occur (i.e., the reception of reflected replicas of the same satellite signal). The accuracy and the precision of the position observations is thus degraded. In this project the student will investigate collaborative navigation scenarios in which mutliple drones measure their relative positions via computer vision and attempt to cope with local GNSS unavailability or degraded performance. There are many possible aspects to be explored depending on the candidate’s own preferences/background and available credits. Examples follow:

  • develop real-time LEDs tracking software (computer vision/machine vision cameras/embedded systems/C++)
  • develop state estimation for cooperative localization (pose graph, similar to SLAM, motion models, least-squares estimation, real-time, C/C++)
  • develop algorithms and control mechanisms for formation flight (distributed systems, ROS, autopilots, C/C++/embedded systems)
  • system integration (hardware, linux, networking, cables&soldering, flight control tuning & settings)

Recommended type of project: Master thesis / Semester project (master students)
Work breakdown: 10% theory, 50% software, 40% experimentation
Prerequisites: knowledge of C/C++ and/or Matlab, passion for robotics, love for fliying things.
Keywords: UAVs, robotics, autonomous navigation, simulation
Contacts: Davide A. Cucci


Data Fusion Methods for Integrated Navigation

INS+GNSS position error

Navigation systems for outdoor applications rely mainly on satellite positioning (GNSS) and inertial navigation (IMU). The latter system is indispensable for attitude determination and for providing the autonomy in frequent cases when the environment disturbs the reception of satellite signals. For applications where a low cost, size and weight are important, MEMS-IMUs are currently the only option. This is the case, for instance, with mini and micro Autonomous Aerial Vehicles (UAV). Inherently, the low accuracy of MEMS-IMUs makes the autonomous determination of the trajectory challenging in many aspects. This project focuses on comparing several methods of data fusion when integrating GNSS receivers with inertial sensors. Some novel approaches on autonomous navigation were recently developed in our lab (TOPO), testing of which will be part of this project. Apart from simulation-based testing (on an internal soft-platform), real (life) tests are necessary to assess the performances of each navigation system and/or method. Such tests can be performed on terrestrial or airborne vehicles (i.e. a car, a helicopter or fixed-wing plane) together with high accuracy sensors (available in the lab) that serve as a reference. While MA semester project maybe limited to (Monte-Carlo) simulations, the Master project is expected to include also real data testing and evaluation.

Recommended type of project: Master thesis / Semester project (master students)
Work breakdown: 20% theory, 30% software, 30% hardware, 20% experimentation
Prerequisites: familiarity with data fusion methods and dynamic systems, good knowledge of MATLAB and C/C++, interest in hardware implementation and experimental jobs (for Master project)
Keywords: integrated navigation, INS/GNSS, data fusion, Kalman filter
Contacts: Mehran Khaghani


Enhancing Aerial Images via Super-Resolution

Super-resolution is a technique that allows to produce higher resolution images out of multiple lowresolution images of the same scene taken from slightly different points of view. In aerial photogrammetry such set of low resolution images can be acquired employing a high frame rate camera and taking advantage of the natural movement of the mapping plaftorm (e.g., a drone). One of the main steps of super-resolution algorithms is to deterimine the exact relative view-points from which the images were taken (i.e., image registration). This is normally done via matching the same visual features on different images. Alternatively, or in addition, inertial information coming from accelerometers and gyroscopes can be employed, if they are available. The goal of this project is to experiment super-resolution algorithms in aerial-mapping applications.

super resolution principle
An overview of the super-resolution principle,
from “Super-Resolution Image Reconstruction: a Technical Overview”, 
IEEE signal processing magazine, 2003

Depending on the number of credits the student will concentrate on one or more of the following aspets:

  • implement one super-resolution algorithm based on existing methods based on matching of visual features,
  • perfect the camera acquisition software available at TOPO to acquire burst of images very close in time,
  • employ inertial readings to improve the relative view-points determination and compare with visual-only algorithms
  • design lab experiments to precisely evaluate the geometric correctness of the produced super-resolution images

Recommended type of project: Master thesis / Semester project (master students)
Work breakdown: 20% theory, 50% software, 30% experimentation
Prerequisites: knowledge of C/C++ and/or Matlab, understanding of signal processing, some image processing experinece is reccommended.
Keywords: image processing, UAVs, inertial navigation
Contacts: Davide A. Cucci


Autonomous UAV Vehicle Following
Based on Optical Targets

realm world vehicle following experiments
PX4 SITL Simulation based on ROS and Gazebo Terrain vehicle following experiment ad EPFL.

A new mapping paradigm, which is currently being investigated by TOPO, is based on a terrestrial vehicle (TV)/micro aerial vehicle (UAV) tandem. In this setup, the UAV follows autonomously the TV and augments laser-scanning readings taken from the ground with aerial images. One of the possible ways to achieve the following is by tracking the optical target present on the top of the TV. In this master/semester project the student will investigate autonomous navigation and control strategies for the UAV employing the Software-In-the-Loop simulator developed for the PX4 autopilot. This project builds on vast background material, tools and experiments already available at the lab. Some of the objectives of the project are:

  • develop a control architecture so that the UAV follows the ground vehicle always maintaining the optical target in the center of the camera image.
  • the PX4 autopilot can be controlled in different modes, e.g., local level velocity, absolute position, orientation, etc. Experiment with different control variables, architectures and parameter tuning to optimize the following performances.
  • develop target re-acquisition strategies in case the optical target is lost.

Recommended type of project: Master thesis / Semester project (master students)
Work breakdown: 20% theory, 50% software, 30% experimentation
Prerequisites: knowledge of C/C++ and/or Matlab, good understanding of control theory
Keywords: UAVs, robotics, autonomous navigation, simulation
Contacts: Davide A. Cucci


Visual Odometry 
in Micro Aerial Vehicles Orientation Determination

A demo of the SVO: Semi-Direct Visual Odometry package, from UTZ.

Recently we have seen the spread of autonomous/tele-operated micro aerial vehicles (UAVs) both in commercial and hobby application. As the diffusion of such systems increases, so does the demand for safety and robust operation in non-optimal conditions, such as when flying in GNSS restricted areas. One of the current trend in UAVs research focuses on enriching and diversifying the information sources for position and orientation determination, for example introducing tracking cameras and visual odometry/simultaneous localization and mapping (SLAM) algorithms. In this project the student will explore the wide variety of open-source software packages and experiment them on custom UAVs available at TOPO. The specific goal of this project may vary according to the student preferences:

  • develop the hardware and software components necessary to couple the visual odometry/sensor fusion system with the PX4 autopilot and test it in real flights.
  • explores the possibilities offered by the availability of a real-time rough 3D terrain model in autonomous navigation for aerial photogrammetry applications, for instance terrain following or obstacle avoidance.
  • implement a loosely coupled integration scheme of visual, inertial and GNSS sensors, by means of a Kalman filter or another sensor fusion mechanism and compare the real-time results with the trajectory determined by means of offline processing.
  • extend a visual odometry system to handle GNSS position feedback, implementing a tightly coupled sensor fusion scheme.

Recommended type of project: Master thesis / Semester project (master students)
Work breakdown: 20% theory, 40% software, 20% hardware, 20% experimentation
Prerequisites: good knowledge of C/C++, interest in computer vision and robotics
Keywords: visual odometry, computer vision, SLAM, UAVs, aerial photogrammetry, robotics
Contacts: Davide A. Cucci


A Dynamic Simulator
for Unmanned Fixed Wing Micro Aerial Vehicles

a view of the last_letter simulation package
(a view of the last_letter simulation package)

UAV simulation in Gazebo

At TOPO we employ micro aerial vehicles (UAVs), both commercial and internally developed, for aerial photogrammetry. As more advanced applications and autonomous behaviors are developed, it becomes critical to run testing in an accurate and realistic simulation environment, before actual flights take place. After a preliminar analysis of the requirements, the student will review the available software packages, then build a fixed wind micro aerial vehicle simulator based on one of the available open-source robotic simulation frameworks (e.g., Gazebo). The goals of the project can be tuned according to the specific skills and inclinations of the student. Some of these are:

  • review and eventually extend the aerodynamic models employed in the physical simulation. Compare them with the ones developed at TOPO for UAV position and attitude determination.
  • determine how to tune the developed simulator to target a specific, real world, UAV, as the ones available at TOPO. Compare the simulation results with data captured during real flights. Explore the performance of the simulator in handling limit situations such as low altitude flights (e.g., during landing) and in presence of strong wind.
  • implement a realistic simulation of a camera mounted on the simulated UAV. Render the images given a 3D terrain model or an orthophoto. 
  • review, investigate and eventually extend the hardware-in-the-loop simulation capabilities, when the simulator is coupled with the PX4 open-source autopilot.

Recommended type of project: Master thesis / Semester project (master students)
Work breakdown: 20% theory, 40% software, 10% hardware, 30% experimentation
Prerequisites: familiarity with physical modeling and motion equations, good knowledge of C/C++, broad interest in robotics, love for flying things
Keywords: simulation, fixed wing UAVs, aerial photogrammetry, robotics
Contacts: Davide A. Cucci


If you have interest in this domain, please don’t hesitate to contact TOPO staff.

Visit IS-Academia project page for specific proposals.