Vision for Robotics LabOpen OpportunitiesGraph 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.
- Computer Vision, Intelligent Robotics, Robotics and Mechatronics
- Semester Project
| 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. - Computer Vision, Image Processing, Intelligent Robotics, Signal Processing
- Master Thesis, Semester Project
| The project aims to implement a semantic label transfer from satellite to aerial imagery in order to enable the training of image-based machine learning algorithms for autonomous aerial vehicle tasks, such as path planning, collision avoidance, and localization. - Computer Vision, Intelligent Robotics
- Semester Project
|
|
