This workshop will be held at the 14th IEEE International Conference on Automation Science and Engineering (CASE 2018) in Munich, Germany.
National Institute of Standards and Technology (NIST) – Intelligent Systems Division
E-mail: firstname.lastname@example.org (Phone: +1 301-975-3455)
National Institute of Standards and Technology (NIST) – Intelligent Systems Division
E-mail: email@example.com (Phone: +1 301-975-3510)
DLR German Aerospace Center – Institute of Robotics and Mechatronics
E-mail: firstname.lastname@example.org (Phone: +49 8153 28-1175)
A new class of robots called collaborative robots or Co-Bots are designed to safely work alongside human workers in manufacturing environments. These robots are equipped with force sensing and/or compliance through series elastic actuator technologies in order to limit forces and prevent injury to humans working in their proximity. Next generation end-effector technologies such as fully-actuated and under-actuated robotic hands with advanced force control, in-hand manipulation capabilities and built in compliance are mainstream within the research community. These new technologies as well as machine learning/artificial intelligence techniques and simplified programming interfaces show promise as new ways for tackling the small parts assembly problem of trending low-volume, high-mix production operations.
This workshop will help identify the challenges associated with implementing robotic assembly by first exploring the application space through presentations by leading experts in the automotive, aerospace, and consumer goods manufacturing sectors. In addition, keynote presentations and a poster session will highlight work underway that is already addressing the challenges in this application space. Finally, tools to benchmark research progress in mechanical assembly that are designed with reference to existing design-for-assembly (DFA) methods will be presented. The discussion period will be used to identify key research areas needed to address the low-volume, high-mix, small parts assembly problem. Although the focus will be on assembly applications, many of the same challenges and solutions are relevant to other domains, such as service robotics.
Topics of interest include:
Call for Posters
We kindly invite you to submit contributions to a poster session at this workshop that will give the opportunity to researchers to discuss their latest results and ongoing research activities with the community. During a fast talk presentation, presenters will have one minute to introduce their work to all workshop attendees.
To participate, please submit by June 15 2018, an extended abstract of 1-2 pages (in PDF format, IEEE regular paper format) via email to all the organizers:
– Joe Falco (email@example.com)
– Elena Messina (firstname.lastname@example.org)
– Maximo A. Roa (email@example.com)
All contributions will undergo a review by the organizers, and the authors will be notified of acceptance by June 30th, 2018.
Korbinian Nottensteiner (DLR, German Aerospace Center, Germany)
Title: Autonomous assembly for one-of-a-kind production
Abstract: This talk presents a fully automated system for automatic assembly of aluminum profile constructions using a dual arm robotic system. The system includes an assembly sequence planner integrated with a grasp planning tool, a knowledge-based reasoning method, a skill-based code generation, and an error tolerant execution engine. The modular structure of the system allows its adaptation to new products, which can prove especially useful for SMEs producing small lot sizes. The system is robust and stable, as demonstrated with the repeated execution of different geometric assemblies
Karl Van Wyk (NIST, National Institute of Standards and Technology, USA)
Title: Replicable Tests and Benchmarking for Robotic Assembly Operations
Abstract: The ability to quantify and compare the performance of robot systems in assembly operations is critical to its development. Researchers in the field must be able to positively identify true advancement, or lack thereof, in their pursuit of improving robot mechanics, sensing, and control for assembly. The National Institute of Standards and Technology (NIST) is developing a series of task boards that target the performance measurement of specific assembly operations. The design of these task boards references existing design-for-assembly research that has methodically quantified various factors that influence human performance in assembly. Low-cost and internationally acquirable, the task board designs are easily replicated to facilitate benchmarking among researchers. Alongside the task boards, NIST has proposed associated test methods, metrics, and statistical tools for rigorously assessing robot assembly performance. As an illustrative example, various robot systems were programmed to complete a task board. The performance of each system was measured and the results were statistically compared to identify true differences in performance data.
Venkat Krovi (Clemson University - Center for Automotive Research, USA)
Title: Learning from Demonstrations for Human-Robot Collaborative Assembly Tasks
Abstract: Human-robot collaborative assembly consists of humans and automated robots, who cooperate with each other to accomplish complex assembly tasks which are difficult for either humans or robots to accomplish alone. There has been some success in statistics-based and optimization-based approaches to achieve human-robot collaboration tasks. In this paper, we take a different approach by introducing convolutional neural networks (CNN) into the teaching-and-learning process in the collaborative assembly tasks. Robots are trained by the dataset online established by human demonstrations. In the training, robots learn the stations of an unknown multi-stage co-assembly task and the dynamical environment of the shared workspace. The trained robots generate correct behaviors to assist humans in collaborative assembly by given real-time images of the shared workspace. In our experiment, we verified our approach by involving the mechanical assembly of a vehicle model on our collaborative assembly test platform. Our results suggest that the dataset for training, validation, and testing can be established within 3-4 demonstrations. The trained robot can intelligently move to next station which is convenient for the human to accomplish the upcoming sub-assembly maneuvers. The trained robot can stop its maneuver immediately whenever human’s hands approach the risk area based on its awareness on the dynamical shared workspace.
Sami Haddadin / Lars Johannsmeier (Franka Emika / Munich School of Robotics and Machine Intelligence, TUM, Germany)
Title: A Control and Learning Framework for Force-Sensitive Robot Manipulation Skills
Abstract: In my talk I will introduce modern control and observer techniques for complex robot systems that physically interact with the world and manipulate it, including various applications in assembly. After shortly reviewing essential robot modeling foundations I introduce the overall problem of interaction control and monitoring, discussing well known concepts such as impedance and force control, arriving then at modern schemes such as energy-based and adaptive control algorithms. Finally, I outline the design and use of nonlinear observers in interaction and collision handling.
Max Reynolds (Symbio Robotic, USA)
Title: AI-based Control Architectures for High-Precision Assembly
Abstract: Developing automated assembly solutions can be challenging in the context of process variation. We present a distributed control architecture for AI-based assembly that reduces the burden on programmers and staff. This architecture leverages a multi-tier computation and control pipeline running on both OT and IT resources. An analysis of performance and scalability of the system are included.
Juan Aparicio Ojea (Siemens, USA)
Title: Robot Learning Benchmarks: insights for comparing robotic assembly performance in ML/AI-based approaches.
Abstract: A dominant trend in manufacturing is the move toward small production volumes and high product variability. It is thus anticipated that future manufacturing automation systems will be characterized by a high degree of autonomy, and will be able to learn new behaviors without explicit programming. Robot Learning, and more generic, Autonomous Manufacturing, is an exciting research field at the intersection of Machine Learning and Automation. The combination of "traditional" control techniques with Artificial Intelligence holds the promise of allowing robots to learn new behaviors through experience. This has motivated many labs around the world to focus their attention to this area of research. The question arises: how can we benchmark different machine learning algorithms and apply them to the challenges of industrial automation? How can we ensure that algorithms are benchmarked in a realistic and fair way? In this talk, we will introduce a benchmark example designed by Siemens researchers (https://www.siemens.com/us/en/home/company/fairs-events/robot-learning…)
Philip Freeman (Boeing, USA)
Title: Title: Opportunities and Challenges in developing and deploying truly collaborative robots
Abstract: The Boeing Company is pushing our production to new levels of automation and robotics in commercial aircraft production, space and defense products, and service and support of existing platforms. We see automation as a key tool for the mechanic to help make their job easier, safer, and more productive. Because of the scope and scale of our manufacturing and service business, collaborative robotics plays an integral part of our automation strategy – both the current generation of shared workspace robotics, as well as helping to develop a future generation of truly collaborative work partners. In this presentation I show some examples of where and how Boeing is using shared workspace robotics, present some of the near term challenges to developing and deploying collaborative robotics in aerospace, and where we see opportunity in advancing the technology to allow people and robots to work side-by-side with the robot assisting the worker in a complex manufacturing environment.
Fan Dai (ABB, Germany)
Title: Thoughts on Robot Assembly Skills and Machine Learning
Abstract: Industrial robots are designed and programmed to do desired tasks proficiently, efficiently, robust, and reliably. With the rising demand on flexibility of production and ease of use of robot systems, the requirements on the skills of the robot systems are significantly increasing. There are many approaches of implementing robot assembly skills and also various approaches for robot skill learning since decades. In this talk, we describe our view on this very interesting and demanding topic with some samples from our investigations and test implementations. Potentials of applying machine learning to obtain or enhance robot skills are discussed.
Yiannis Karayiannidis (Chalmers University of Technology, Sweden)
Title: Robotic assembly with bimanual and collaborative robots under uncertainty
Abstract: The SARAFun project (H2020, Mar 2015- Feb 2018) moved beyond traditional industrial assembly applications where uni-manual robotic setups are considered by employing dual arm human-sized robots to perform assembly tasks that are typically executed by humans. In this talk I will focus on dual arm robot control strategies employed in SARAFUN for assembly application scenarios beyond peg-in-hole, such as folding and insertion by deformation. I will discuss how dual arm collaborative robots can show increased flexibility and independence from a structured environment. To this end, I will present an interactive perception framework based on adaptive control, kinesthetic sensing (force/torque) and proprioception. I will also outline human-in-the-loop approaches for cooperative robotic assembly.
9:00 – 10:30 Invited Speaker Presentations
10:30 – 11:00 Coffee Break
11:00 – 12:00 Invited Speaker Presentations
12:00 – 12:30 Flash Poster Presentations
12:30 – 2:00 Poster Sessions / Lunch
2:00 – 3:30 Invited Speaker Presentations
3:30 – 4:00 Coffee Break (Posters)
4:00 – 4:30 Invited Speakers Presentations
4:30 – 5:30 Final Discussion
Posters (Extended Abstracts):