BaCalibrationVisualization

Chair for Computer Aided Medical Procedures & Augmented Reality
Lehrstuhl für Informatikanwendungen in der Medizin & Augmented Reality

THIS WEBPAGE IS DEPRECATED - please visit our new website

User supporting Visualisationmethods for Hand-Eye Calibration

Abstract

In this project the student should think about various methods to support common, but yet complex calibration tasks in the the field of Augmented Reality. In Augmented Reality scenarios a user/developer typically needs to undergo several calibration tasks depending on the type of sensors to build an AR application . Most famous calibration tasks would be:
  • Camera Intrinsics Calibration (for all vision based systems)
  • Absolute Orientation (combining sensor information from same coordinate system)
  • Hand-Eye Calibration (combining sensors from distinct coordinate systems)
  • Tip Calibration (Calibration for pointing devices)
  • Room Calibration (for stationary vision based systems)
  • SPAAM Calibration (Calibration of HMDs)

Usually all solutions of these calibration methods are estimated using numerical solvers in the first place before perfoming some non-linear optimization for an accurate result. Depending on the linear system or applied solver (SVD, Eigenvalues, etc.) the results of these estimations are always influenced by the input data, typically gathered from the user/developer of the AR system. Unexperienced users can easily gather lots of data which is not suitable for the underlaying calibration process and usually achieve only disapointing results in the calibration process which also affects the AR applications. It usually takes lots of time/trials for new users/developers in the field of Augmented Reality to be aware of common mistakes in these AR calibration scenarios.
The goal of this thesis is to build a system that supports a common calibration procedures via an adequate online visualization framework. The online system should be build such that it perceives sensor information via websockets and that uses the sensor information to display clear instructions to the user to achieve on optimal calibration movements/results. This should support a better estimation in at least one of the above listed common calibration methods and avoid the frustration of unexperienced AR users/developers in complex sensor setups.

Available resources during the project

  • Dependings on the choosen calibration problem, hardware (access to the lab) will be provided to the students for testing the framework

Student tasks (What you will have to do)

  • Literature research on the choosen calibration procedure to investigate optimal type of input data
  • Literature research in 3DUIin general, and User Interfaces
  • Documentary research of the needed tools and technology (Self-Introduction in the used technology)
  • Time scheduling of the whole thesis
  • Implementation of self-explanatory visualization methods/metaphors that supoort an optimal calibration procedure.
  • Write the study design and study protocol
  • Evaluate the implementation in an exhaustive user study to achieve results meeting requirements of a scientific publication
  • Analyze evaluation results
  • Write your thesis (in English or German)

Student requirements (what you should already know in the best place)

  • Unity 3D experience or another 3D (gaming) environment of his choice
  • Basic knowledge of 3D Modelling (Polygons, Meshes, Textures, etc.)
  • Knowledge of JSON & Websocket technology

Contacts


ProjectForm
Title: User supporting Visualisationmethods for Hand-Eye Calibration
Abstract:  
Student: Michael Boxhammer
Director: Gudrun Klinker
Supervisor: Christian Waechter
Type: DA/MA/BA
Area: Industrial Tracking, Computer Vision, Industrial Augmented Reality, Medical Augmented Reality
Status: finished
Start:  
Finish:  
Thesis (optional):  
Picture:  


Edit | Attach | Refresh | Diffs | More | Revision r1.2 - 02 Jan 2019 - 11:42 - TobiasLasser