Summary:
The overarching goal of the RoLei research project is to enable robotic systems to install deformable cables, which are used, for example, in the hydraulic systems of agricultural machinery. By automating this previously manual assembly task, manufacturing companies can reduce waste, relieve employees from monotonous and non-ergonomic tasks, and increase the efficiency of cable installation. To achieve these goals, RoLei explores a combined approach of cable simulation and a real robotic system. In the first step, boundary conditions for gripper design and robot path planning are derived from the cable simulation. The latter is then adapted to the real robotic system. Additionally, novel approaches for the subsequent control of installed cables are being developed.
Possible tasks:
Depending on the student’s interests and capabilities, the following tasks could be agreed upon:
One potential area of focus is the development of a concept for robot-based quality control using multiple sensors. This involves designing and integrating sensor systems to ensure the precise and reliable installation of deformable cables, enhancing overall quality assurance.
Another task is the design and construction of use case-specific test rigs. Students can work on creating experimental setups tailored to specific scenarios, enabling thorough testing and validation of robotic cable installation processes.
Additionally, students can contribute by developing a concept for robotic grippers in confined spaces. This involves designing grippers that can operate efficiently in limited spaces, a crucial aspect for applications in complex machinery. A prototype of the gripper can be produced using additive manufacturing technologies and subsequently tested.
Implementing subprocesses of robot-based cable assembly is another key area. Students can work on integrating and optimizing various subprocesses involved in the robotic installation of cables, ensuring seamless and efficient operations.
Students can focus on improving robot vision using AI. This involves leveraging advanced machine learning models to enhance the robot’s visual perception, enabling more accurate and adaptive cable handling. The goal would be to achieve instance segmentation in video sequences leveraging state-of-the-art models such as Segment Anything 2.
Furthermore, students can work on the development of an algorithm for the iterative optimization of the cable installation trajectory, considering the deformation of the cable. This task involves creating and refining algorithms that adapt the robot’s path in real-time to account for the flexible nature of the cables, ensuring precise and efficient installation.
Another critical area is bridging the simulation-reality gap by linking the cable installation trajectory with the robot kinematics. This involves developing methods to seamlessly transfer simulation results to the real-world robotic system, ensuring that the planned trajectories are accurately executed by the robot, thus overcoming discrepancies between simulated and actual performance.
Learning outcomes:
The learning outcomes associated with the completion of specific tasks may vary. Depending on the content of the tasks, the following skills may be developed during the course of work:
- Programming in Python or C++ (e.g. in ROS)
- Developing and testing algorithms
- Understanding robotics and automation
- Designing and 3D printing components
- Working with sensors and 3D cameras
- Applying computer vision techniques
- Training AI models
- Simulating robotic movements
- Using containerization (e.g., Docker)
