Student Projects


How to apply

To apply, please send your CV, your Ms and Bs transcripts by email to all the contacts indicated below the project description. Do not apply on SiROP . Since Prof. Davide Scaramuzza is affiliated with ETH, there is no organizational overhead for ETH students. Custom projects are occasionally available. If you would like to do a project with us but could not find an advertized project that suits you, please contact Prof. Davide Scaramuzza directly to ask for a tailored project (sdavide at ifi.uzh.ch).


Upon successful completion of a project in our lab, students may also have the opportunity to get an internship at one of our numerous industrial and academic partners worldwide (e.g., NASA/JPL, University of Pennsylvania, UCLA, MIT, Stanford, ...).



Fast Monocular Visual-inertial Odometry - Available

Description: Monocular visual-inertial (VI) system is the minimal setup that can provide reliable state estimation for robotics. Among different monocular VI odometry algorithms, optimization-based methods has become popular because of its superior accuracy. However, most of them are too expensive to run in real-time on an embedded system. The goal of this project is to integrate an optimization backend with our fast visual odometry frontend SVO to achieve reliable state estimation using only onboard sensing and computing. The applicant is required to be proficient in C++ programming and have computer vision knowledge.

Goal: The project aims to develop fast visual-inertial odometry algorithms that can execute in real-time on embedded systems.

Contact Details: Zichao Zhang (zzhang at ifi.uzh.ch)

Thesis Type: Semester project / Master Thesis

See project on SiROP

Online loop detection and closing - Available

Description: Loop detection and closing technique help robots recognize visited places. It can be used to recover odometry failure as well as reduce drift. However, it is still challenging to run such algorithms in real-time on an embedded system. The goal of this project is to integrate a loop closing and detection algorithm into our visual odometry frontend. We aim to run the loop detection and closing algorithm using only onboard computing resource and enable our robot to localize reliably. The applicant is required to be proficient in C++ programming and have computer vision knowledge.

Goal: The project aims to develop loop detection and closing algorithms suitable for embedded systems.

Contact Details: Zichao Zhang (zzhang at ifi.uzh.ch) Titus Cieslewski (titus at ifi.uzh.ch)

Thesis Type: Semester project / Master Thesis

See project on SiROP

Study the Motion Limit of Visual Odometry - Available

Description: Visual odometry estimates the motion of the camera based on the images. It is an important algorithm in robotics and becomes widely used nowadays. However, when putting visual odometry into real world robotic applications, we need to understand the limitations of the algorithm. This project aims to address the question: how fast can a robot move while still keeping the visual odometry working properly? It can be expected that different visual odometry algorithms have different tolerance for fast motion. Therefore, such knowledge will enable designing a robust algorithm that switches between different visual odometry methods depending on the camera motion.

Goal: Specific work will include the theoretical analysis of the visual odometry pipeline and validation by simulation/experiments. It is also possible to perform the study for different camera configurations.

Contact Details: Zichao Zhang (zzhang at ifi.uzh.ch)

Thesis Type: Semester project / Master Thesis

See project on SiROP

Robust and Adaptive Multi-Camera Visual Odometry - Available

Description: Most visual odometry algorithms are designed to work with monocular cameras and/or stereo cameras. One way to improve the robustness of visual odometry is to use more cameras (3 to N). While it is relatively easy to make visual odometry work with multiple cameras for a specific type of configuration, developing an adaptive solution that works with arbitrary camera configurations (i.e., without changing the code) and that is robust to failures (i.e., if one camera fails during the execution, the algorithm can still proceed) is not straightforward.

Goal: The project aims to develop a robust and adaptive multi-camera visual odometry pipeline based on the existing framework in our lab.

Contact Details: Zichao Zhang (zzhang at ifi.uzh.ch)

Thesis Type: Semester project / Master Thesis

See project on SiROP

Smart Feature Selection In Visual Odometry - Available

Description: For most robotic platforms, computational resources are usually limited. Therefore, ideally, algorithms running onboard should be adaptive to the available computational power. For visual odometry, the number of features largely decides the resource the algorithm needs. By using a selected subset of features, we can reduce the required computational resource without losing accuracy significantly.

Goal: The project aims to study the problem of smart feature selection for visual odometry. The student is expected to study how motion estimation is affected by feature selection (e.g., number of features, different feature locations). The ultimate goal will be to implement a smart feature selection mechanism in our visual odometry framework.

Contact Details: Zichao Zhang (zzhang at ifi.uzh.ch)

Thesis Type: Semester project / Master Thesis

See project on SiROP

Benchmarking camera control for visual odometry - Available

Description: There are many existing datasets to evaluate the performance of visual odometry algorithms. However, few work has been done in providing a principled way to benchmark the performance of camera control (exposure time/gain) algorithms, which have a large impact on the performance of visual odometry. Most of the current datasets/benchmark tools simply contain images captured at a certain camera configuration, which is not suitable for this purpose. A proper benchmark tool can fill this gap and will be useful for understanding the strengths and weaknesses of different algorithms.

Goal: The goal of this project is to make use of both synthetic and real data to build a benchmark tool and evaluate the influence of different camera control algorithms on the performance of visual odometry.

Contact Details: Zichao Zhang (zzhang at ifi.uzh.ch)

Thesis Type: Semester project / Master Thesis

See project on SiROP

Online time offset estimation for visual-inertial system - Available

Description: Visual-inertial odometry (VIO) has progressed significantly recently and finds a lot of real-world applications. One of the crucial requirement for good performance is to have a synchronized camera and inertial measurement unit. However, many low-cost systems do not have good synchronization, which limits the use of VIO. As an alternative, the time offset can be estimated by software. Existing methods to estimate the time offset either operate offline or only applies to specific algorithms. A lightweight algorithm that can estimate the camera-IMU offset will greatly extend the application scenarios of VIO.

Goal: The goal of the project is to develop an efficient and flexible algorithm to estimate the time offset between a camera and an IMU.

Contact Details: Zichao Zhang (zzhang at ifi.uzh.ch)

Thesis Type: Semester project / Master Thesis

See project on SiROP

Motion-aware camera control - Available

Description: It is well know that the camera needs to be set to proper exposure time and gain to work well in practice. In the case of visual odometry, however, the motion of the camera also needs to be considered. For example, if the exposure time is too high, too much motion blur will also cause the image to degrade. Therefore, the camera control algorithm should also take into consideration of the camera motion to optimize the performance of visual odometry.

Goal: The goal of this project is to develop a camera control algorithm that is aware of the camera motion and can adjust the exposure time considering both motion blur and the scene brightness.

Contact Details: Zichao Zhang (zzhang at ifi.uzh.ch)

Thesis Type: Semester project / Master Thesis

See project on SiROP

Decentralized Visual Place Recognition in the Real World - Available

Description: We have recently developed decentralized multi-robot visual place recognition (neural network based) and SLAM and demonstrated them on well-known datasets ( http://rpg.ifi.uzh.ch/docs/arXiv17_Cieslewski.pdf ). We want to take this work one step further and deploy it in the real world with a group of quadrotors. This is where you come in.

Goal: In this project, you will analyse the feasibility of real-world decentralized visual SLAM from one key aspect: Visual Place Recognition. You will help plan the field experiment by engineering for performance and robustness. Based on data you collect, you will provide design decisions for various aspects of the experiment. These will be up to you (what are the main bottlenecks?), but examples include: Camera placement on the robot, active camera control, neural network fine-tuning, where to do the experiments (default option is our offices), adapting the environment with minimal effort, …

Contact Details: Titus Cieslewski ( titus at ifi.uzh.ch ), appreciated skills: Linux, ROS, good programming skills, ideally in Matlab or Python

Thesis Type: Semester project / Master Thesis

See project on SiROP

Communication for a (Real) Group of Robots - Available

Description: One of our visions for the future is to deploy a group of robots in an unknown environment and have them create a map of that environment. However, a key obstacle to doing this in the real world is communication. Although there are many theoretical solutions, it is a very common theme in the community that deploying them in practice is hard.

Goal: In this work, you will approach this problem from a strictly practical perspective. You will operate on the concrete case of an experiment where a group of flying robots is to be deployed in the offices of our lab. You will start with the simple approach of a well-placed router, and evaluate what we can afford in that scenario, and what factors impact reliability. If that approach is well-explored, you will consider more advanced approaches such as one of the robots providing a wifi hotspot for the other robots, or using specialized hardware like Zigbee. An interesting sub-project would be to create a communication map ( https://tinyurl.com/y7g5qmfc ) of the offices.

Contact Details: Titus Cieslewski ( titus at ifi.uzh.ch ), required skills: Linux, ability to learn autonomously, a sense for the practical.

Thesis Type: Semester project / Master Thesis

See project on SiROP

Simulating decentralized multi-robot SLAM - Available

Description: We have recently developed decentralized multi-robot visual place recognition (neural network based) and SLAM and demonstrated them on well-known datasets ( http://rpg.ifi.uzh.ch/docs/arXiv17_Cieslewski.pdf ). We want to take this work one step further and deploy it in the real world with a group of quadrotors. Since this is quite an effort, logistically, a first step will be to simulate the full system (SLAM, but also obstacle avoidance and control) in simulation.

Goal: In this project, you will simulate a scenario where a group of quadrotors explores and maps an unknown environment. We will start with a simplistic simulation and gradually increase its complexity. Axes in which to complexity can be increased: From manual camera placement to using a full control stack, from rendering camera frames in Gazebo to rendering them in a more photorealistic simulator, from random motion with reactive obstacle avoidance to active exploration, from a few robots to many robots, …

Contact Details: Titus Cieslewski ( titus at ifi.uzh.ch ) Required skills: Linux, experience in ROS or a very strong ability to learn, C++/Python.

Thesis Type: Semester project / Master Thesis

See project on SiROP

Trajectory estimation and scene reconstruction from any YouTube video! - Available

Description: We believe that with the right processing, it should be possible to obtain trajectory estimation and scene reconstruction from many of the videos that are already out there on the internet! This could have nice applications: two approaches that we would be quite passionate about would be a) to visualize FPV races ( https://www.youtube.com/watch?v=EcLk_uZe33w ) and b), more practical for the community, to create new robotics datasets with little effort.

Goal: Go from YouTube videos to trajectory estimation and scene reconstruction. The more general, the better. Of course, you will start with an as simple as possible approach: Slow motion, 360° videos (theoretically no need to calibrate), then increase in complexity. While this will likely be rigged with engineering, hacks and qualitative evaluation, it might, with the right approach, also lead to interesting research (e.g. how to deal with motion blur in tracking, auto-calibration, potential to apply machine learning etc...).

Contact Details: Titus Cieslewski ( titus at ifi.uzh.ch ), appreciated skills: Linux, ROS, good programming skills, ideally in Matlab or Python. Prefer students who took the Vision Algorithms for Mobile Robots class!

Thesis Type: Semester project / Master Thesis

See project on SiROP

Visual Bundle Adjustment with an Event Camera - Available

Description: Event cameras such as the Dynamic Vision Sensor (DVS) are recent sensors with large potential for high-speed and high dynamic range robotic applications. The goal of this project is to improve an existing visual odometry pipeline using an event camera by designing and integrating a visual bundle adjustment module in order to reduce the drift in the odometry pipeline. A good theoretical background on computer vision is necessary to undertake this project. The candidates will be expected to be comfortable with C++ as well.

Contact Details: Henri Rebecq (rebecq at ifi.uzh.ch), Guillermo Gallego (guillermo.gallego at ifi.uzh.ch)

Thesis Type: Master Thesis

See project on SiROP

A real-time Event Camera Simulator - Available

Description: Event cameras such as the Dynamic Vision Sensor (DVS) are recent sensors with large potential for high-speed and high dynamic range robotic applications. Recently, a variety of new algorithms to perform various vision tasks, such as fast object detection, visual odometry, or depth estimation using this sensor have emerged. However, the performance of these methods are usually not extensively assessed, for lack of ground truth data. To fill this gap, the goal of this project is to extend and improve an existing event camera simulator developed in our lab.

Goal: The two major objectives of this project are: 1. integrate a real-time rendering engine (raw OpenGL, Unity, Unreal Engine ?) to our simulator in order to provide a real-time simulated event stream 1. implement a realistic Inertial Measurement Unit (IMU) simulator.

Contact Details: The expected candidate for this project should have a background in computer graphics, and be comfortable with programming with C++. Previous experience with 3D software or rendering engines such as Unity or Unreal Engine would be a must. Henri Rebecq (rebecq at ifi.uzh.ch)

Thesis Type: Semester project / Bachelor Thesis

See project on SiROP

A 3D reconstruction algorithm using a stereo pair of event cameras - Available

Description: Event cameras such as the Dynamic Vision Sensor (DVS) are recent sensors with large potential for high-speed and high dynamic range robotic applications.

Goal: The goal of this project is to use a stereo pair of event cameras to obtain a 3D reconstruction of a scene. The student will extend a recent event-based 3D reconstruction approach developed by our lab for monocular event cameras to the case of a pair of stereo event cameras.

Contact Details: Applicants should have a good background in computer vision (especially stereo reconstruction techniques), and should be comfortable with C++. Henri Rebecq (rebecq at ifi.uzh.ch)

Thesis Type: Semester project / Master Thesis

See project on SiROP

Optical Flow Estimation with an Event Camera - Available

Description: Event cameras such as the Dynamic Vision Sensor (DVS) are recent sensors with large potential for high-speed and high dynamic range robotic applications. The goal of this project is to use event cameras to compute the optical flow in the image plane induced by either a moving camera in a scene or by moving objects with respect to a static event camera. Several existing methods as well as proposed new ones will be analyzed, implemented and compared. A successful candidate is expected to be familiar with state-of-the-art optical flow methods for standard cameras. This is a project with considerable room for creativity, for example in applying the ideas from low-level vision or ideas driving optical flow methods for standard cameras to the new paradigm of event-based vision. Experience in coding image processing algorithms in C++ is required.

Contact Details: Guillermo Gallego (guillermo.gallego at ifi.uzh.ch), Henri Rebecq (rebecq at ifi.uzh.ch)

Thesis Type: Master Thesis

See project on SiROP

Building a high-speed camera! Learning Image reconstruction with an Event Camera - Available

Description: Event cameras such as the Dynamic Vision Sensor (DVS) are recent sensors with large potential for high-speed and high dynamic range robotic applications. The output of an event camera is a sparse stream of events that encode only light intensity changes - in other terms, a highly compressed version of the visual signal.

Goal: The goal of this project is to turn an event camera into a high-speed camera, by designing an algorithm to recover images from the compressed event stream. Inspired by a recent approach, the goal of this project will be to train a machine learning algorithm (or neural network) to learn how to reconstruct an image from the noisy event stream. The first part of the project will consist in acquiring training data, using both simulation and real event cameras. The second part will consist in designing and training a suitable machine learning algorithm to solve the problem. Finally, the algorithm will be compared against state-of-the-art image reconstruction algorithms. The expected candidate should have some background on both machine learning and computer vision (or image processing) in order to undertake this project.

Contact Details: Henri Rebecq (rebecq at ifi.uzh.ch), Guillermo Gallego (guillermo.gallego at ifi.uzh.ch)

Thesis Type: Master Thesis

See project on SiROP

A Visual-Inertial Odometry System for Event-based Vision Sensor - Available

Description: Event-based cameras are recent revolutionary sensors with large potential for high-speed and low-powered robotic applications. The goal of this project is to develop visual-inertial pipeline for the Dynamic and Active Vision Sensor (DAVIS). The system will estimate the pose of the DAVIS using the event stream and IMU measurements delivered by the sensor. Filtering approaches as well as batch optimization methods will be investigated. https://youtu.be/bYqD2qZJlxE http://www.inilabs.com/products/davis

Contact Details: Henri Rebecq (rebecq at ifi.uzh.ch), Guillermo Gallego (guillermo.gallego at ifi.uzh.ch)

Thesis Type: Master Thesis

See project on SiROP

Continuous Structure From Motion (SfM) with an Event Camera - Available

Description: Event cameras such as the Dynamic Vision Sensor (DVS) are recent sensors with large potential for high-speed and high dynamic range robotic applications. The goal of this project is to explore the possibilities that continuous structure from motion (SfM) has to offer for event cameras. In the continuous formulation, visual egomotion methods attempt to estimate camera motion and scene parameters (depth of 3D points) from observed local image velocities such as optical flow. This formulation is appropriate for small-baseline displacements, which is the scale at which events are fired by a moving DVS in a static scene. Several ideas from classical and new methods will be taken into account to address the challenges that the fundamentally different output of a DVS poses to the SfM problem. The expected candidate should have good theoretical as well as programming skills to undertake this project.

Contact Details: Guillermo Gallego (guillermo.gallego at ifi.uzh.ch), Henri Rebecq (rebecq at ifi.uzh.ch)

Thesis Type: Master Thesis

See project on SiROP

Precision landing for a cargo drone - Available

Description: Multicopters are recently used to deliver different types of goods in dense environments such as cities, due to their manoeuvrability and possibility for vertical take-off and landing. However, landing among high buildings may cause reflections or even loss of the GPS signal causing imprecision in the localisation of the drone, thus unprecise landing. In this project, we aim at using a vision based technique to land precisely in a designated place by a drone sender or recipient. Specifically, the first goal of this project will be to localise and remember a take-off position of a drone for precise landing in the same place after delivery. The second task is to find a landing spot indicated on Google maps by a recipient. Thirdly, the algorithm should verify the position of obstacles on the ground such as people, cars, trees etc. and not land on them. Then, the landing performance will be validated on the prototype of the safe foldable delivery drone developed at EPFL.

Goal: The goal of this project is to design and implement a vision based solution for precise landing in cluttered outdoor environments. This project will be done in collaboration between two labs: Robotic and Perception Group from the University of Zurich, and Laboratory of Intelligent Systems from Ecole Polytechnique de Lausanne.

Contact Details: Davide Scaramuzza (sdavide@ifi.uzh.ch), Przemyslaw Kornatowski (przemyslaw.kornatowski@epfl.ch)

Thesis Type: Master Thesis

See project on SiROP

3rd Person View Imitation Learning - Available

Description: Manually programming robots to carry out specific tasks is a difficult and time consuming process. A possible solution to this problem is to use _imitation learning_, in which a robot aims to imitate a teacher, e.g., a human, that knows how to perform the task. Usually, the teacher and the learner share the same point of view on the problem. However, this last assumption might not be necessary. As humans, for example, we learn to cook by looking at others cooking. During this project, we will explore the possibility of repeating such a kind of 3rd person view _imitation learning_ with flying robots on a navigation task.

Goal: The project aims to develop machine learning based techniques that will enable a drone to learn flying by looking at an other robot flying.

Contact Details: **Antonio Loquercio**: loquercio@ifi.uzh.ch

Thesis Type: Semester project / Bachelor Thesis / Master Thesis

See project on SiROP

Safe Reinforcement Learning for Robotics - Available

Description: Reinforcement Learning (RL) has recently emerged has a technique to let robots learn by their own experience. Current methods for RL are very data-intensive, and require a robot to fail many times before actually accomplishing their goal. However some systems, such as flying robots, require to respect safety constraints during learning and/or deployment. While maximizing performance, those methods usually aim to minimize the number of system failures and overall risk.

Goal: During this project, we will develop machine learning based techniques to let a (real) drone learn to fly nimbly through gaps and gates, while minimizing the risk of critical failures and collisions.

Contact Details: **Antonio Loquercio** loquercio@ifi.uzh.ch

Thesis Type: Semester project / Master Thesis

See project on SiROP

Simulation to Real World Transfer - Available

Description: Recent techniques based on machine learning enabled robotics system to perform many difficult tasks, such as manipulation or navigation. Those techniques are usually very data-intensive, and require simulators to generate enough training data. However, a system only trained in simulation (usually) fails when deployed in the real world. In this project, we will develop techniques to maximally transfer knowledge from simulation to the real world, and apply them to real robotics systems.

Goal: The project aims to develop techniques based on machine learning to have maximal knowledge transfer between simulated and real world on a navigation task.

Contact Details: **Antonio Loquercio** loquercio@ifi.uzh.ch

Thesis Type: Semester project / Bachelor Thesis / Master Thesis

See project on SiROP

Optical Flow-Based Obstacle Avoidance - Available

Description: Optical flow describes the motion in a picture plane, which can be computed by algorithms from sequences of normal images or even event cameras. The optical flow holds interesting information for MAVs like quadrotors, since it makes approaching obstacles detectable. It becomes especially interesting when executing agile, fast maneuvers, where a reactive obstacle avoidance is needed. This enables safe and fast navigation and might possibly even reduce the trajectory planning effort needed. To enable reactive obstacle avoidance, this projects first targets to implement an optical flow algorithm or use an existing one. Then a simple control strategy to avoid approaching obstacles should be developed and finally tested in simulation as well as in real world experiments. An optional extension might be to leverage the information from optical flow and use it in short time horizon planning to avoid obstacles efficiently while following a trajectory in a optimal control fashion.

Goal: - Search existing optical flow algorithms and/or implement one with the focus on fast execution. - Implement and evaluate control policies based on the optical flow to avoid obstacles. - Optional: Leverage the exploitation of this information to path planning methods based on optimization.

Contact Details: Philipp Föhn (foehn at ifi.uzh.ch)

Thesis Type: Semester project / Master Thesis

See project on SiROP

Realization of an High-Speed Flight Experimental Setup based on a Tether - Available

Description: Quadrotors are great vehicles for fast and agile maneuvers and new control hardware and software allows to exploit such fast flying situations. Generally, the modelling of such quadrotors ignores aerodynamic effects or simplifies them, but there are also more complex models that can compensate for such effects. This project would aim at a setup where a quadrotor is fixed to a point with a tether and flies fast circles at the limit of this tether (like in https://youtu.be/iJPy1geXu4M).This way, the centrifugal forces are captured by the tether and the quadrotor thrust can fully be used for acceleration and fast flight. This setup is capable of high speed flights in a confined environment and could be used to identify aerodynamic model parameters at high velocities, resulting in better precision and simple data gathering.

Goal: - Some mechanical work building the tether setup. - Implementing a controller to fly the tethered cirlces. - Identify an approximation of the aerodynamic model and the parameters.

Contact Details: Philipp Föhn (foehn at ifi.uzh.ch)

Thesis Type: Semester project / Master Thesis

See project on SiROP

Implementation of a Model Predictive Control Algorithm with Extension to Obstacle Avoidance - Available

Description: To control a quadrotor in very fast, dynamic and agile maneuvers, it is important to have reliable controllers running at high speed onboard the quadrotor, which can also take into account nonlinearities and predict behaviour over a short time horizon. Model Predictive Control (MPC) algorithms cover this need by (re-)generating trajectories of a short future time horizon that are optimal to some cost with respect to some reference. They offer a very powerful and versatile way to control quadrotors even in complex environments and fast maneuvers. This project therefore aims at an implementation of such an MPC controller for trajectory tracking and navigation based on our existing state-dependent LQR controllers or with a new approach. This will be tested on a new, small state-of-the-art quadrotor platform and optionally extended with obstacle avoidance, for example based on a stereo camera.

Goal: - Implement and test (simulation and real) an iterative LQR algorithm. - Make an efficient and fast implementation on our new small quadrotor. - Optional: Extend the approach to obstacle avoidance.

Contact Details: Philipp Föhn (foehn at ifi.uzh.ch)

Thesis Type: Semester project / Master Thesis

See project on SiROP

Hand-Eye Calibration Toolbox - Available

Description: Hand-Eye calibration is a paramount pre-processing stage of many robotic and augmented reality applications, where the knowledge of the relative transformation between different sensors (e.g. a camera and a head-mounted display) is required to have an accurate geometric representation of the scene.

Goal: The goal of this project is to develop a user-friendly hand-eye calibration toolbox integrated with our robotic system. The toolbox will contain existing and novel hand-eye calibration methods, and it will allow to visualize the results of the different methods in an integrated manner to improve the understanding of the quality of the processed dataset, specially paying attention to error estimates, uncertainties and detection of inconsistent data.

Contact Details: Guillermo Gallego (guillermo.gallego at ifi.uzh.ch)

Thesis Type: Semester project / Master Thesis

See project on SiROP

Integrated Multi-Camera Calibration Toolbox - Available

Description: The toolbox is expected to handle different camera brands, projection models and calibration patterns. In the multi-sensor scenario, the toolbox is also expected to compute the temporal offsets between the sensors. Special attention will be given to estimation of error measures, parameter uncertainties, detection of inconsistent data and interactive guidance of data acquisition.

Goal: The goal of this project is to develop a user-friendly, single and multi-camera calibration toolbox adapted to our robotic system. The toolbox will integrate existing calibration software in our group and in other libraries and will provide user-friendly reports of the different stages to assess the quality of the processed dataset, thus speeding up and improving the understanding of the whole sensor calibration stage.

Contact Details: Guillermo Gallego (guillermo.gallego at ifi.uzh.ch)

Thesis Type: Semester project / Master Thesis

See project on SiROP

Robust Control for Quadrotors - Available

Description: While a quadrotor is in flight, it may experience disturbances caused by wind gust, rotor drag, interaction with environment, and so on. Since those disturbances can affect the flight quality and accuracy significantly, they should be well estimated and compensated. In this project, we aim at designing a robust quadrotor controller which can reject external disturbances and enhance overall quadrotor control accuracy. Specifically, the first goal of this project will be estimating disturbances by exploiting various state feedbacks. After estimating the disturbances, in the quadrotor controller, the estimated disturbance values will be reflected and compensated. To ensure the stability of the robust quadrotor controller, first, a relevant stability analysis will be conducted. Then, the flight performance under disturbances will be validated on an agile quadrotor platform.

Goal: The goal of this work is designing and implementing a robust quadrotor controller to deal with external forces / torques.

Contact Details: Suseong Kim (suseong@ifi.uzh.ch), Davide Falanga (falanga@ifi.uzh.ch)

Thesis Type: Semester project / Master Thesis

See project on SiROP

Fast Minimum Time Trajectory Generation for Quadrotors - Available

Description: Quadrotors have been utilized as a versatile airborne platform. Especially, it has been applied to monitor, reconnaissance, and surveillance tasks by exploiting its various flight regimes from hovering to aggressive maneuvering. To utilize a quadrotor more effectively, in this project, we aim at designing quadrotor trajectories to fly through the predefined waypoints in the time optimal manner. In the trajectory generating considerations, we mainly focus on two aspects: (i) designing feasible trajectories in terms of the given actuator capacity; (ii) computing trajectories in real time to adapt the reference trajectory in accordance with dynamic environments. The computed trajectory will be compared with the existing minimum time trajectory generating methods. Also, experiments using an agile quadrotor will be conducted to validate the feasibility of the computed trajectory.

Goal: The goal of this work is to design a minimum time trajectory generation algithrhm. Specifically, this project aims at computing a feasible minimum time trajectory in real time.

Contact Details: Suseong Kim (suseong@ifi.uzh.ch), Davide Falanga (falanga@ifi.uzh.ch)

Thesis Type: Semester project / Master Thesis

See project on SiROP

Collision Free Trajectory Generation - Available

Description: To safely operate a quadrotor in cluttered environments, a quadrotor should be able to avoid obstacles while flying. It can be done by observing the environment around a quadrotor and designing a reference trajectory that does not collide with detected obstacles. In this project, we aim at designing a collision free trajectory generation method using visual information provided by an onboard monocular or stereo camera. The visual information will be used to extract non-occupied space around the quadrotor. Then, we can generate the reference trajectory within the computed free space. To do that, an optimization problem will be formulated to minimize some criteria such as actuation values, jerk, snap, or the distance to the goal while constraining the distance between the quadrotor and occupied regions. The trajectory generation method will be tested in the simulation environment first. After that, it will be evaluated with an agile quadrotor platform in indoor and outdoor environments.

Goal: The goal of this work is designing a collision free trajectory using visual information from an onboard monocular or stereo camera.

Contact Details: Suseong Kim (suseong@ifi.uzh.ch), Davide Falanga (falanga@ifi.uzh.ch)

Thesis Type: Semester project / Master Thesis

See project on SiROP

Adaptive Trajectory Sampling and Control Throttling for High-Speed Flight - Available

Description: Numerous solutions to plan obstacle-free, high-speed trajectories for quadrotors have been proposed in the last years, and state-of-the-art nonlinear controller can be used to track them. One of the issue arising from a digital implementation of a controller tracking an high-speed trajectory is related to the frequency the latter should be sampled. The goal of this project is to analyze the impact of the sampling frequency on the tracking performance of a quadrotor flying an high-speed trajectory, and to design an adaptive sampling scheme taking into account the "aggressiveness" of such a trajectory to help the controller tracking it more accurately. Also, such an adaptive scheme will be used to throttle the controller according to such the sampling frequency.

Goal: The goal of this project is to analyze the impact of the sampling frequency on the tracking performance of a quadrotor flying an high-speed trajectory.

Contact Details: Davide Falanga (falanga@ifi.uzh.ch), Suseong Kim (suseong@ifi.uzh.ch)

Thesis Type: Semester project / Master Thesis

See project on SiROP

Obstacle Avoidance with Quadrotors Using a Neuromorphic Event Camera - Available

Description: Event cameras such as the Dynamic Vision Sensor (DVS) are recent sensors with large potential for high-speed and high dynamic range robotic applications. One of the main advantages of such sensors is the reduced latency when compared to standard cameras, which makes them particularly suited for tasks where latency is critical, such for examples obstacle avoidance with flying vehicles. This projects aims at implementing and evaluating state-of-the-art algorithms for time to contact using a Neuromorphic Event Camera. Such algorithms can be applied to obstacle avoidance with quadrotors, in order to let the vehicle safely navigate without hitting unexpected obstacles along its path.

Goal: This projects aims at implementing and evaluating state-of-the-art algorithms for time to contact using a Neuromorphic Event Camera. Such algorithms can be applied to obstacle avoidance with quadrotors.

Contact Details: Davide Falanga (falanga@ifi.uzh.ch)

Thesis Type: Semester project / Master Thesis

See project on SiROP

Modelling, Control and Planning for a Shape-Shifting Quadrotor - Available

Description: Recent works have demonstrated that micro quadrotors are extremely agile and versatile vehicles, able to execute very complex maneuvers. However, the majority of the quadrotors available on the market rely on a fixed mechanical structure, which cannot be changed while flying. The goal of this project is to enable quadrotors to change their morphology while they are airborne, guaranteeing stability during the entire flight, including the switching phase. More specifically, we are interested in quadrotors able to rotate their arms in order to obtain a configuration of the actuation system that best fits the task the vehicle has to execute. For example, is some applications it might be required that the vehicle spans the smallest volume possible, while in others performances might have higher priority. In this project, the student will have to: (i) derive a dynamical model a quadrotor with switching morphology; (ii) design a feedback controller able to stabilize independently on the current configuration, also during the switching phase; (iii) exploit the aforementioned model to plan trajectories that fulfill the system dynamics. The results will be validated in simulation first, and in a second stage on a real quadrotor platform.

Goal: This projects aim at studying the behavior of a quadrotor able to change its morphology while flying. More specifically, the goal is to derive a mathematical model capturing its dynamics and design a control law able to stabilize it.

Contact Details: Davide Falanga (falanga@ifi.uzh.ch), Suseong Kim (suseong@ifi.uzh.ch)

Thesis Type: Semester project / Master Thesis

See project on SiROP

Visual Localization with Event Cameras - Available

Description: Visual localization is the problem of estimating the pose of a camera, i.e., the camera position and orientation, with respect to a 3D model of a scene. Solving this problem is a key step for many interesting applications such as autonomous vehicles, e.g., self-driving cars or drones, and Augmented Reality. The typical approach to solving this problem is to establish 2D-3D matches between features extracted from an images and 3D points in a pre-built map of the scene. These 2D-3D matches are then used to estimate the camera pose. The goal of this master thesis is to develop and implement an algorithm for visual localization for event-based cameras, i.e., cameras that only report changes in pixel intensities rather than the original intensities. This algorithm will make use of novel features derived from events that will be developed as part of the thesis.

Goal: Develop and implement an algorithm for visual localization for event-based cameras. This algorithm will make use of novel features derived from events that will be developed as part of the thesis.

Contact Details: Torsten Sattler torsten.sattler@inf.ethz.ch (main contact) Davide Scaramuzza sdavide@ifi.uzh.ch

Thesis Type: Master Thesis

See project on SiROP

Physical Threshold Detection - Available

Description: In human environments, windows and doors represent thresholds between spaces. For a robot exploring an unknown environment, these portals offer new frontiers, but they can be challenging for a robot to safely traverse. This project deals with designing a system to robustly detect windows/doors/thresholds that a quadrotor could fly through, using cameras and range sensors. This system would need to detect these portals via appearance and geometry, and evaluate the feasibility of traversal with a very low false-positive rate using machine learning techniques and geometric constraints. The ultimate goal is to deploy this system on a quadrotor for a live demo.

Contact Details: Jeff Delmerico (jeffdelmerico at ifi.uzh.ch)

Thesis Type: Semester project / Master Thesis

See project on SiROP