Fraunhofer Institute for Factory Operation and Automation IFF

Hand to eye calibration is one of the most important steps to realize high flexibility and digitization of industry. In this paper, base on the research background of improving the degree of automation and flexibility in the field of aircraft fuselage inspection and detection, the hand to eye calibration under large size of operating volume is firstly studied based on indoor global positioning system (iGPS). A new calibration mathematical model A _i x = Yb _i is firstly proposed under this research background, and the solution of this model is given based on the orthogonal matrix property and multi-objective optimization algorithm includes the Particle Swarm Algorithm (PSO) and Difference Evolutionary Algorithm (DE). Verification experiments have shown that the method can achieve 3D camera calibration with an average error around 0.1021 mm in a large workspace. Thus, it is proved through experiments that this method can realize high-precision and high-robustness calibration of 3D camera, tool center point (TCP) coordinate, and tool axis vector in a large scale of operating environment.

Numerical simulation has great potential to provide a more comprehensive understanding of human impact response to injury mechanisms. Finite Element (FE) models are used as a tool to study human injuries in greater detail, for example, the THUMS (Total Human Model Safety of TOYOTA) model, which is widely used as a reliable human model in different fields to predict human injuries such as fractures, internal organ damage, and brain tissue injuries. However, no available FE model can be used to simulate human‐robot collisions based on standards ISO/TS 15066 and the biomechanical characteristics of human soft tissues in vivo. The authors have developed a head model based on the structures (dimensions and anatomy) of the THUMS head model, specifically designed to simulate impact loads on the masticatory muscles. Based on medical imaging (MRI) data, the soft tissues at the location of the masticatory muscles in the THUMS head are transformed from monolayer to multilayer, that is, a composite geometry of skin‐fat‐muscle each with its own material model and parameters. The model was optimized and validated using the experimental data from the Fraunhofer IFF subjects study, which determined biomechanical thresholds for specific body locations in ISO/TS 15066 under dynamic collisions.

The point cloud is one of the measurement results of local measurement and is widely used because of its high measurement accuracy, high data density, and low environmental impact. However, since point cloud data from a single measurement are generally small in spatial extent, it is necessary to accurately globalize the local point cloud to measure large components. In this paper, the method of using an iGPS (indoor Global Positioning System) as an external measurement device to realize high-accuracy globalization of local point cloud data is proposed. Two calibration models are also discussed for different application scenarios. Verification experiments prove that the average calibration errors of these two calibration models are 0.12 mm and 0.17 mm, respectively. The proposed method can maintain calibration precision in a large spatial range (about 10 m × 10 m × 5 m), which is of high value for engineering applications.

The Fraunhofer IFF has developed a patented model (Fast Response Model = FRM) for safe Human‐Robot Collaboration (HRC), enabling the calculation of reaction forces in the event of a collision between a contact body of any shape and a selected human body part. This paper specifically focuses on adapting the model for five selected body locations on the hand‐arm system. A comparative analysis is conducted, contrasting the results obtained from the FRM with those derived from existing finite element reference models to assess the plausibility of the model's system response. Additionally, computational times of the model are compared with those of established models for simulation‐based risk evaluation of robot collisions. The model's quality is validated using a validation geometry. Results indicate that depending on the selected body part, end effector geometry, and given displacement, similar results to a finite element model can be obtained. However, simulations with the FRM are less time‐consuming. In the future, the data‐driven model could be transformed into a real‐time model using machine learning techniques.

As of today (2022), approximately 51% of electrical energy in Germany is provided from renewable energy sources. The remaining 49% is generated by the conversion of primary energy carriers such as coal, gas, oil, etc. The phase-out of nuclear energy has just been completed in Germany, so there are no longer any nuclear power plants connected to the grid.

Throughout human history, the transition to new methods of energy production (horsepower—steam power—electricity) has led to leaps in human development and the associated improvement in quality of life. However, humanity has never been as dependent on the use of energy as it is today. Without a modern energy system, no industrial nation is conceivable.

As already mentioned, the energy transition is a global task and can succeed as such if all states participate appropriately. The EU has had the energy transition on its agenda for years. The FitFor55 program (55% emission reduction by 2030) and the decision for Net-Zero, i.e. emission neutrality by 2050, are strong signals for other regions of the world.

The parallel machine scheduling problem (PMSP) involves the optimized assignment of a set of jobs to a collection of parallel machines, which is a proper formulation for the modern manufacturing environment. Deep reinforcement learning (DRL) has been widely employed to solve PMSP. However, the majority of existing DRL-based frameworks still suffer from generalizability and scalability. More specifically, the state and action design still heavily rely on human efforts. To bridge these gaps, we propose a practical reinforcement learning-based framework to tackle a PMSP with new job arrivals and family setup constraints. We design a variable-length state matrix containing full job and machine information. This enables the DRL agent to autonomously extract features from raw data and make decisions with a global perspective. To efficiently process this novel state matrix, we elaborately modify a Transformer model to represent the DRL agent. By integrating the modified Transformer model to represent the DRL agent, a novel state representation can be effectively leveraged. This innovative DRL framework offers a high-quality and robust solution that significantly reduces the reliance on manual effort traditionally required in scheduling tasks. In the numerical experiment, the stability of the proposed agent during training is first demonstrated. Then we compare this trained agent on 192 instances with several existing approaches, namely a DRL-based approach, a metaheuristic algorithm, and a dispatching rule. The extensive experimental results demonstrate the scalability of our approach and its effectiveness across a variety of scheduling scenarios. Conclusively, our approach can thus solve the scheduling problems with high efficiency and flexibility, paving the way for application of DRL in solving complex and dynamic scheduling problems.

The decarbonization potential of hydrogen offers increasing usage paths in the fight against climate change resulting in a growing demand for climate‐neutral hydrogen. This challenge is met by producing hydrogen microbially from renewable substrates as an alternative to ‘green hydrogen’ from water electrolysis. Initial results have shown that coupling dark fermentation and anaerobic digestion is not only possible but also advantageous. Specifically, by integrating dark fermentation in existing biogas plants, the overall physical efficiency of the process's substrate turnover can be increased by up to 50% through providing hydrogen in addition to biogas. The achieved test results are examined based on limit‐oriented physical efficiency evaluation to show the potential for optimization of the substrate turnover in biological concepts based on modeling. Finally an overview of a commissioned demonstration plant is given, which will provide further insights into the feasibility of the dark fermentation on an industrial scale.

2D X-ray images are extensively employed for intraoperative navigation and localization owing to their high imaging efficiency, low radiation risk, and affordability. However, this method can only yield overlapped anatomical information from a restricted number of projected views. Conversely, intraoperative CT scanning techniques, offering 3D images, elevate the risk of radiation exposure for both patients and healthcare professionals. For this purpose, we propose a V-shaped convolutional attention mechanism network (X-CTCANet) designed for X-ray reconstruction of CT images. The network enhances reconstruction performance by promoting task consistency in encoding–decoding, minimizing semantic differences between feature mappings. Additionally, it introduces an adaptive convolutional channel attention mechanism to compel the network to prioritize essential feature regions. Experimental results demonstrate the successful CT image reconstruction from spine X-rays using X-CTCANet, achieving an SSIM value of 0.805 and a PSNR value of 34.64 dB. This underscores the considerable potential of accurate 3D CT reconstruction from 2D X-ray images in offering image support for surgical robots.

Green chemistry aims to use renewable materials, reduce waste, and avoid toxic substances and is defined by 12 principles of green chemistry. Green metrics (GM) are measurable figures that assess adherence to the 12 principles of green chemistry. GM are designed to be user‐friendly and can be applied without detailed process knowledge. Life cycle assessment (LCA) is another approach used to estimate the environmental impacts of products or processes throughout their life cycle. This paper compares LCA and GM, exploring their suitability for assessing the greenness of chemical products and processes. This includes the discussion of strengths and weaknesses, limitations, application areas, benefits of combining both approaches, and how to handle conflicting results.

High-precision tracking and localization of the end of industrial robots is one of the most important technologies to realize high efficiency, high precision, and high safety of robots. iGPS (indoor Global Positioning System), as a widely used indoor measurement system, has many advantages that cannot be replaced by other measurement devices, such as high scalability, simultaneous measurement of multiple targets, etc. However, due to the unique measurement principle of iGPS, it does not have good dynamic measurement performance. In order to enhance the dynamic measurement performance of iGPS and maximize its advantages, this paper proposes a correction algorithm that can effectively improve the dynamic measurement accuracy of iGPS based on the ideas of "transforming motion into static" and "equivalent substitution" after in-depth research on the measurement principle of iGPS. To verify the correctness of the mathematical principle of the algorithm and the value of engineering application, this paper has carried out a MATLAB simulation test and real measurement test. The experimental results show that the proposed correction algorithm can effectively improve the dynamic measurement accuracy of iGPS, and the correction algorithm can reduce the global error of the measurement results to within 0.1mm in any robot trajectory and reasonable speed range, which is in line with the general requirements of industrial measurement.

2D X-ray images are extensively employed for intraoperative navigation and localization owing to their high imaging efficiency, low radiation risk, and affordability. However, this method can only yield overlapped anatomical information from a restricted number of projected views. Conversely, intraoperative CT scanning techniques, offering 3D images, elevate the risk of radiation exposure for both patients and healthcare professionals. For this purpose, we propose a V-shaped convolutional attention mechanism network (X-CTCANet) designed for X-ray reconstruction of CT images. The network enhances reconstruction performance by promoting task consistency in encoding-decoding, minimizing semantic differences between feature mappings. Additionally, it introduces an adaptive convolutional channel attention (CCA) mechanism to compel the network to prioritize essential feature regions. Experimental results demonstrate the successful CT image reconstruction from spine X-rays using X-CTCANet, achieving an SSIM value of 0.805 and a PSNR value of 34.64 dB. This underscores the considerable potential of accurate 3D CT reconstruction from 2D X-ray images in offering image support for surgical robots.

The usage of lidar sensors is becoming increasingly more popular in a variety of industrial applications. Retroreflective markers can be used to support the analysis of the lidar data by providing highly reflective landmarks in the generated point clouds. After giving a short overview over lidar and the principle behind retroreflective markers our work concentrates on the real-world application of these markers and the benefits and challenges, they provide. We give a comprehensive overview over existing applications and their used types of retroreflective markers. We also present three of our own use cases that demonstrate the benefits retroreflective markers can provide for the analysis of lidar data in industrial contexts. A lidar based solution for crane operations surveillance in a foundry is presented that uses retroreflective markers to simplify object localisation processes and increases their stability. A second application uses markers to support the AI based classification of hard to distinguish object categories. Lastly the usage of markers in an underground mining scenario for SLAM is presented. We as well point out the considerations that have to be taken into account before choosing a specific retroreflective marker and using it in a real-world application.

This interdisciplinary study examined the relationship between bone density and drilling forces required during trans-pedicular access to the vertebra using fresh–frozen thoraco-lumbar vertebrae from two female body donors (A, B). Before and after biomechanical examination, samples underwent high-resolution CT-quantification of total bone density followed by software-based evaluation and processing. CT density measurements (n = 4818) were calculated as gray values (GV), which were highest in T12 for both subjects (GVmaxA = 3483.24, GVmaxB = 3160.33). Trans-pedicular drilling forces F (Newton N) were highest in L3 (FmaxB = 5.67 N) and L4 (FmaxA = 5.65 N). In 12 out of 13 specimens, GVs significantly (p < 0.001) correlated with force measurements. Among these, Spearman correlations r were poor in two lumbar vertebrae, fair in five specimens, and moderately strong in another five specimens, and highest for T11 (rA = 0.721) and L5 (rB = 0.690). Our results indicate that CT-based analysis of vertebral bone density acquired in anatomical specimens is a promising approach to predict the drilling force appearance as surrogate parameter of its biomechanical properties by e.g., linear regression analysis. The study may be of value as basis for biomechanical investigations to improve planning of the optimal trajectory and to define safety margins for drilling forces during robotic-assisted trans-pedicular interventions on the spine in the future.