Student Projects
LiDAR-Visual-Inertial Odometry with a Unified Representation
Lidar-Visual-Inertial odometry approaches [1-3] aim to overcome the limitations of the individual sensing modalities by estimating a pose from heterogenous measurements. Lidar-inertial odometry often diverges in environments with degenerate geometric structures and visual-inertial odometry can diverge in environments with uniform texture. Many existing lidar-visual-inertial odometry approaches use independent lidar-inertial and visual-inertial pipelines [2-3] to compute odometry estimates that are combined in a joint optimisation to obtain a single pose estimate. These approaches are able to obtain a robust pose estimate in degenerate environments but often underperform lidar-inertial or visual-inertial methods in non-degenerate scenarios due to the complexity of maintaining and combining odometry estimates from multiple representations.
Keywords
Odometry, SLAM, Sensor Fusion
Labels
Semester Project , Master Thesis
Description
Work Packages
Requirements
Contact Details
More information
Published since: 2024-12-16 , Earliest start: 2025-01-05 , Latest end: 2025-06-30
Applications limited to ETH Zurich
Organization Autonomous Systems Lab
Hosts Mascaro Rubén , Chli Margarita
Topics Information, Computing and Communication Sciences
Odometry and Mapping in Dynamic Environments
Existing lidar-inertial odometry approaches (e.g., FAST-LIO2 [1]) are capable of providing sufficiently accurate pose estimation in structured environments to capture high quality 3D maps of static structures in real-time. However, the presence of dynamic objects in an environment can reduce the accuracy of the odometry estimate and produce noisy artifacts in the captured 3D map. Existing approaches to handling dynamic objects [2-4] focus on detecting and filtering them from the captured 3D map but typically operate independently from the odometry pipeline, which means that the dynamic filtering does not improve the pose estimation accuracy.
Keywords
Odometry, Mapping, SLAM, Dynamic Environments
Labels
Semester Project , Master Thesis
Description
Work Packages
Requirements
Contact Details
More information
Published since: 2024-12-16 , Earliest start: 2025-01-05 , Latest end: 2025-06-30
Applications limited to ETH Zurich
Organization Autonomous Systems Lab
Hosts Mascaro Rubén , Chli Margarita
Topics Information, Computing and Communication Sciences , Engineering and Technology
Multi-Sensor Semantic Odometry
Semantic segmentation augments visual information from cameras or geometric information from LiDARs by classifying what objects are present in a scene. Fusing this semantic information with visual or geometric sensor data can improve the odometry estimate of a robot moving through the scene. Uni-modal semantic odometry approaches using camera images or LiDAR point clouds have been shown to outperform traditional single-sensor approaches. However, multi-sensor odometry approaches typically provide more robust estimation in degenerate environments.
Keywords
Odometry, Sensor fusion, Semantics
Labels
Semester Project , Master Thesis
Description
Work Packages
Requirements
Contact Details
More information
Published since: 2024-12-09 , Earliest start: 2024-07-14 , Latest end: 2025-01-31
Applications limited to ETH Zurich
Organization Autonomous Systems Lab
Hosts Chli Margarita , Mascaro Rubén
Topics Information, Computing and Communication Sciences
Dense Monocular SLAM with 3D Gaussians
This project offers a unique opportunity to work on cutting-edge technologies shaping the future of AR/VR and robotic perception. The selected candidate will benefit from direct supervision by researchers in Federico Tombari’s team at Google.
Keywords
Simultaneous Localization and Mapping, SLAM, 3D Gaussian Splatting, 3DGS
Labels
Master Thesis
Description
Work Packages
Requirements
Contact Details
More information
Published since: 2024-12-09 , Earliest start: 2025-01-06 , Latest end: 2025-06-30
Applications limited to ETH Zurich
Organization Autonomous Systems Lab
Hosts Mascaro Rubén , Chli Margarita
Topics Information, Computing and Communication Sciences
Data-efficient learning of world models
Model-based reinforcement learning uses world models to predict the outcomes of a robot's actions, allowing it to "imagine" new interaction trajectories and reduce data needs. However, learning a high-fidelity world model requires a large amount of data. This thesis aims to minimize the amount of real-world data required to learn a world model. We will staet experimenting with simple tasks, such as pushing objects on a table, and eventually test our ideas on a real robot arm.
Keywords
World models; Deep Learning; Reinforcement Learning; Robotics
Labels
Master Thesis
Description
Work Packages
Requirements
Contact Details
More information
Published since: 2024-12-06 , Earliest start: 2025-01-01 , Latest end: 2025-09-30
Applications limited to ETH Zurich
Organization Autonomous Systems Lab
Hosts Schiavi Giulio
Topics Information, Computing and Communication Sciences , Engineering and Technology
Vision- and Depth-based Reinforcement Learning for dexterous and accurate manipulation with Aerial Robots [multiple openings]
Our team develops novel Aerial Robots that are able to autonomously manipulate and perform work in flight. In this thesis, we would like to explore the learning of task-specific policies for manipulation in flight.
Keywords
Reinforcement Learning, Aerial Robotics, Manipulation, Drones, drone, uav, aerial robot, learning, data-driven
Labels
Master Thesis
Description
Work Packages
Requirements
Contact Details
More information
Published since: 2024-12-03 , Earliest start: 2024-09-30
Applications limited to ETH Zurich , University of Zurich
Organization Autonomous Systems Lab
Hosts Pantic Michael , Allenspach Mike , Stastny Thomas
Topics Information, Computing and Communication Sciences
Gaussian Belief Propagation using Bayes Models
Gaussian Belief Propagation (GBP) has emerged as a promising factor graph inference method, offering remarkable scalability and flexibility, particularly in distributed optimization across multiple agents. However, GBP faces several significant challenges, including convergence issues, the dependency on reliable initial estimates, and the need for an efficient scheduling scheme to coordinate message passing. This thesis project aims to address these challenges by exploring the integration of Bayesian modeling into the GBP process, to improve initialization and scheduling, ultimately enhancing the robustness and effectiveness of GBP in practical applications.
Keywords
Gaussian Belief Propagation (GBP), Factor Graph Optimization, Bayes Net, Bayes Tree
Labels
Master Thesis
More information
Published since: 2024-11-19 , Earliest start: 2024-08-25
Organization Autonomous Systems Lab
Hosts Hug David , Chli Margarita
Topics Mathematical Sciences , Information, Computing and Communication Sciences
Autonomous Mapping with Dynamic Obstacles
Autonomous mapping systems can provide crucial information for agricultural monitoring and industrial inspection tasks. They aim to map unknown environments without human intervention using online mission planning. Many approaches for online mission planning exist but only a few can handle the presence of dynamic obstacles [1-3], which produce moving occlusions and can result in incomplete maps.
Keywords
Drones, Mapping, Planning
Labels
Semester Project , Master Thesis
Description
Work Packages
Requirements
Contact Details
More information
Published since: 2024-11-01 , Earliest start: 2024-01-01 , Latest end: 2024-07-01
Organization Autonomous Systems Lab
Hosts Pinto Teixeira Lucas
Topics Information, Computing and Communication Sciences , Engineering and Technology
Aerial Autonomy in Challenging Dynamic Environments
Automating drone navigation promises to revolutionise the way we conduct a wide variety of tasks, such as agricultural monitoring, industrial inspection, and disaster relief scenarios. Equipping a drone with the capability to autonomously explore and map previously unseen environments using onboard sensors and algorithms forms the basis of autonomy. While there has been tremendous progress in this area over the past few years [1-5], existing systems still lack reliability and adaptability to the challenges and complexity of real settings, which is crucial for the deployment of this technology in actual missions. In particular, performing robust navigation and mapping in highly dynamic environments (e.g., forests) remains an open challenge. Following promising leads from the state-of-the-art and our in-house navigation stack, the goal of this project is to develop the capability to deal with increasingly dynamic and complex scenarios. The student will be guided towards leveraging the multi-sensor capabilities of a LiDAR-Visual-Inertial payload being developed in the lab to research approaches for perception and mission planning that can fuse information from the different sensors and capture high-fidelity representations of challenging dynamic environments. Initially, the student will work within a realistic simulation environment and then deploy and test their work onboard a real drone in a real setting.
Labels
Master Thesis
Work Packages
Requirements
Contact Details
More information
Published since: 2024-11-01 , Earliest start: 2024-02-20 , Latest end: 2024-12-05
Organization Autonomous Systems Lab
Hosts Pinto Teixeira Lucas
Topics Information, Computing and Communication Sciences
Robust Graph Optimization for Robotic Perception
Graph optimization is a key technique employed within Simultaneous Localization And Mapping (SLAM) frameworks, enabling the automation of robot navigation. By encoding a mobile robot’s experiences of the world in a graph (i.e., the robot poses and the sensor readings), such techniques offer a robust way of estimating the robot’s trajectory and the map of its environment. However, these techniques are often computationally demanding and their performance is severely hampered by outliers originating from erroneous sensor measurements or incorrect loop closure detections. To address this challenge, robust graph optimization methods, such as Pairwise Consistency Maximisation (PCM) [2] and Graduated Non-Convexity (GNC) [3], have been devised aiming to enhance the resilience of SLAM algorithms against such outliers, however robustness and complexity both remain open challenges to date. This project aims to build on the strengths of existing works on robust graph optimization in the context of SLAM, and guide the development and implementation of optimization methods robust to outliers caused by erroneous measurements and loop closure detections. Moreover, techniques such as incremental computation will be investigated to better tailor graph optimization to real-time SLAM applications, which is a requirement in robot navigation. This project is offered by the Vision for Robotics Lab (www.v4rl.com) at ETH Zurich and the University of Cyprus. Students undertaking the project may have the opportunity to visit the lab at the University of Cyprus, but this is not required.
Keywords
SLAM; Optimization; Mobile Robots; Computer Vision
Labels
Semester Project
Description
Work Packages
Requirements
Contact Details
More information
Published since: 2024-10-16 , Earliest start: 2024-07-17 , Latest end: 2024-09-30
Organization Autonomous Systems Lab
Hosts Pinto Teixeira Lucas , Chli Margarita
Topics Information, Computing and Communication Sciences , Engineering and Technology
Design of a compliant robotic end-effector with tactile sensing
Unlike robots, humans can easily and reliably grasp objects of arbitrary shapes with their hands, even without seeing them. The goal of this project is to design a compliant robotic end-effector and equip it with tactile sensing to achieve similar capabilities on a robot.
Keywords
Soft robotics, tactile sensing
Labels
Semester Project
PLEASE LOG IN TO SEE DESCRIPTION
More information
Published since: 2024-09-23 , Earliest start: 2024-09-23 , Latest end: 2025-04-12
Applications limited to ETH Zurich
Organization Autonomous Systems Lab
Hosts Ott Lionel , Hüfner Antonia
Topics Engineering and Technology
Note on plagiarism
We would like to suggest every student, irrespective of the type of project (Bachelor, Semester, Master, ...), to make himself/herself familiar with ETH rules regarding plagiarism