Sensor Fusion of Event-Based and Frame-Based Cameras for a CubeSat Mission

Institute

Lehrstuhl für Spacecraft Systems (TUM-ED)

Type

Semester Thesis / Master's Thesis /

Content

 

Description

Event-based cameras are increasingly being investigated for space applications such as star tracking, space situational awareness, and optical navigation. Unlike frame-based sensors, event cameras operate on a pixel-by-pixel basis and respond to changes in relative intensity rather than measuring absolute brightness at fixed time intervals. When the change in brightness at a pixel exceeds a defined threshold, the pixel generates an event. Each event encodes the timestamp of the change, the pixel location, and the polarity, indicating whether the intensity increased or decreased. As a result, event cameras produce a continuous, asynchronous stream of events instead of discrete image frames.

The goal of this thesis is to implement and evaluate the fusion of event-based and conventional frame-based cameras to enhance perception by leveraging the strengths of each sensor type. Event cameras capture brightness changes at high temporal resolution and with low latency, making them ideal for detecting fast-moving objects or operating in high dynamic range scenes. In contrast, frame-based cameras capture full images of the scene, providing detailed visual information, but they are limited by frame rate and prone to motion blur.
The thesis is related the EventSat mission, a 6U CubeSat, developed by the Chair of Spacecraft Systems, featuring an event camera coupled with a Jetson Orin Nano for onboard processing. In addition to the event camera, a frame camera will be integrated, whose power supply and processing chain are entirely separate from the event camera.

Requirements

- Solid programming skills, preferably in Python or C++.
- Basic knowledge of computer vision and image processing.
- Basic CAD skills.
- Interest in sensor data processing and machine learning.
- Ability to work independently and document results clearly.

Possible start

sofort

Contact

Lara Schuberth
lara.schuberthtum.de

Announcement