SA/MA/IDP: Vision-based Detection of Connectors for Bin Picking of Wire Harnesses

Institute

Institut für Werkzeugmaschinen und Betriebswissenschaften (TUM-ED)

Type

Semester Thesis / Master's Thesis /

Content

experimental /  

Description

Motivation

Automating wire harness assembly remains an unsolved challenge in industrial production because of their deformability. Computer vision and tactile sensing are promising enablers for robotic automation of such tasks. Both technologies are investigated at the iwb within the applied research project RoLei. Since wire harnesses are supplied in bins, the question arises how individual connectors can be picked in a targeted manner for subsequent robotic assembly.

Objective

The goal of this thesis is to develop and validate a perception pipeline that detects the connectors of a wire harness lying randomly in a bin and estimates their 6D pose for targeted manipulation. State-of-the-art learning-based and/or model-based methods are to be evaluated and further developed for this purpose. The pipeline will be implemented as a robot skill in ROS 2. Occlusions of the cable branches are a particular challenge. A UR10e robot, a 3D camera and tactile sensors are available. 

Requirements

• prior experience in computer vision

• solid programming skills, ideally Python and ROS 2

• interest in robotics

• structured and independent way of working

Please include your current transcript of records in your application.

Possible start

Contact

Florian Pachler
Room: 1308
florian.pachleriwb.tum.de

Announcement