Research
Our work at IRZ is centered around a number of research initiatives that target some of the most pressing challenges in robotic manufacturing. With these initiatives we proactively define the strategic goals of our efforts, ensuring we can systematically generate impactful, applicable and sustainable results.
Robot Accuracy
This initiative addresses critical issues in robot accuracy in both classical "teach and repeat" and dynamic planning scenarios. Achieving high repeatability is challenging due to thermal deformations. To achieve absolute accuracy, necessary in dynamic planning scenarios, especially in lot-size one application, all the relevant effects have to be addressed such as: Thermal deformations, geometric inaccuracies, deflection under load, joint linearities and backlash. The absence of general solutions to these problems makes industrial robots for a lot of use cases unfeasible as kinematic platforms.
The initiative aims to create a universal robot model capable of handling both revolute and prismatic joints and is able to express all relevant effects, in the static and dynamic case. We target to develop a universal approach which works without case-custom modelling and is applicable by technicians.
Currently, the static problem has been solved for revolute joints, and a separate model has been developed for prismatic joints in static scenarios. Future work will focus on dynamic cases for both joint types, with the ultimate goal of creating a comprehensive model for all robot types and possible combinations to make robots accurate and cost-effective kinematic platforms.
Model Predictive Tool-Path Planning
For many complex industrial processes, tool-path planning cannot be set up correctly the first time. Typically, heuristic planners provide an initial plan that is refined through trial and error during production, which is a very time-consuming process and often no suitable tool path can be found, which lead to serious compromises in part quality.
To address this issue, we are developing ultra-fast, data-driven process models that can be automatically calibrated in-line. These models we then use to calculate optimal toolpaths with respect to specific evaluation metrics and setup constraints, utilizing advanced, and most often custom optimization techniques.
We currently focus on Wire Arc Additive Manufacturing (WAAM) and thermal spraying. In WAAM, we are exploring a reinforcement learning-based planning approach that leverages uncertainty predictions our model provides. In thermal spraying, we are applying nonlinear optimization methods to optimize toolpaths, benefiting from the differentiability of the developed model. In the future, we plan to extend our work to other processes and employ machine learning techniques for generating initial paths, aiming to enable fully automated toolpath generation directly from CAD designs.
Rapid & Autonomous Automation
Current automation approaches are optimized for mass production of identical goods, and the engineering of according systems is a costly and labor-intensive process. Reasons for this are fractured and closed ecosystems, impossibility of the reuse of solutions, and the immense requirements on reliability and cost.
This research aims to provide reusable, universally applicable solutions for core challenges in robot-based automation, mirroring achievements in the software industry. Additionally, we seek to enable fully autonomous, lot-size-one manufacturing systems.
Progress includes hardware-independent process definition, intuitive multi-robot teaching, autonomous workspace exploration, fully autonomous 3D measurement, and a assembly cell concept for software-defined manufacturing. Ongoing research focuses on multi-robot path planning, hard real-time control with Python from local servers, and developing a retrofittable collision detection system.
Robotic Milling
Robotic milling presents an opportunity for cost-effective large-scale machining, with industrial robots being ten times more cost-effective than traditional milling machines for the same workspace. However, the challenge lies in achieving precision and productivity when working with metals, as current capabilities are limited to plastics and wood.
In this initiative, our goal is to revolutionize robotic milling in metals by taking a holistic approach. We are developing innovative concepts that address various aspects simultaneously, including robot stiffness and dynamics, machining processes, tooling, control systems, system-aware tool-path planning, and in-line measurement and correction.
We are currently exploring two approaches: For series production, we implement repetitive learning control, periodically measuring machined parts on a Coordinate Measuring Machine (CMM) to adapt the tool-path. In lot-size one scenarios, we have devised a secondary kinematic system that mitigates robot deflections and oscillations in real-time, enhancing stiffness and dynamics. Our future work aims to benchmark these solutions and develop process models for generating tool paths that optimize working points for the system, ultimately enabling productive and accurate robotic milling in metals.