AUTHORS: A. Ambiehl, S. Garnier, K. Subrin
New method for decoupling the articular stiffness identification: Application to an industrial robot with double encoding system on its 3 first axis
In order to be able to perform complex and arduous tasks, stiffness articular identification of industrial robots is a current approach to predict the deflection under static or dynamic loading. Manufacturers propose new features to take the loading into account and a new generation of industrial robot equiped with double encoding systems are proposed. However, current methods brings some drawbacks when the ratio between the stiffness arm and the wrist one is too high. In this paper, we propose a new approach to take this aspect into account by decoupling the arm identification and the wrist one. We compare then our method regarding two current methods and applied it on this new industrial robot. The results highlight the stability and the quality of the stiffness articular estimation with and without activating the double encoding system. On our data, we are able to take into account 84% of the global deflection.
24-28 Sept. 2017Go to complete version
AUTHORS: E. Ozturk, A. Barrios, C. Sun, S. Rajabi, J. Muñoa
Robotic Assisted Milling for Increased Productivity
Robots’ role in machining is growing as they are being used as machining platforms in increasing number of applications. Moreover, robots have an important role in a new application called, robotic assisted milling, where a robot provides additional support to a workpiece when actual machining is performed by a machine tool. In the paper, alternative methods of support with the robot, i.e. fixed and mobile support are explained. Experimental results show that the robot’s support improves the static and dynamic response of the process. Hence, dimensional errors are minimized and surface quality is improved.
26 April 2018
AUTHORS: L. Gutzeit, A. Fabisch, M. Otto, J. H. Metzen, J. Hansen, F. Kirchner, E. A. Kirchner
The BesMan Learning Platform for Automated Robot Skill Learning
We describe the BesMan learning platform which allows learning robotic manipulation behavior. It is a stand-alone solution which can be combined with different robotic systems and applications. Behavior that is adaptive to task changes and different target platforms can be learned to solve unforeseen challenges and tasks, which can occur during deployment of a robot. The learning platform is composed of components that deal with preprocessing of human demonstrations, segmenting the demonstrated behavior into basic building blocks, imitation, refinement by means of reinforcement learning, and generalization to related tasks. The core components are evaluated in an empirical study with 10 participants with respect to automation level and time requirements. We show that most of the required steps for transferring skills from humans to robots can be automated and all steps can be performed in reasonable time allowing to apply the learning platform on demand.
31 May 2018Go to complete version
AUTHORS: W. Ji, Y. Wang, H. Liu, L. Wang
Interface Architecture Design for Minimum Programming in Human-Robot Collaboration
Many metal components, especially large-sized ones, need to be ground or deburred after turning or milling to improve the surface qualities, which heavily depends on human interventions. Robot arms, combining movable platforms, are applied to reduce the human work. However, robots and human should work together due to the fact that most of the large-sized parts belong to small-batch products, resulting in a large number of programming for operating a robot and movable platform. Targeting the problem, this paper proposes a new interface architecture towards minimum programming in human-robot collaboration. Within the context, a four-layer architecture is designed: user interface, function block (FB), functional modules and hardware. The user interface is associated with use cases. Then, FB, with embedded algorithms and knowledge and driven by events, is to provide a dynamic link to the relevant application interface (APIs) of the functional modules in terms of the case requirements.
13 Mar 2018Go to complete version
AUTHORS: A. Gerasimenko, M. Ritou, V. Godreau, B. Furet
Monitoring of Trimming Operation for Lightweight Composite Structure
Carbon or glass fibre reinforced polymer composites are widely used, notably for lightweight structures, due to their interesting material properties. However, the machining of composites represents a challenge. Their excessive abrasiveness leads to tool wear, which can put workpiece material integrity at risk. In the case of lightweight structure, the monitoring of trimming operation is difficult due to vibrations of the flexible workpiece and scrap offcut. This paper presents a monitoring approach based on vibration and power signals that is compatible with the machining of lightweight composite structure. Tap tests are performed to analyze the experimental set-up and the machining tests. Experiments validate the approach with detections of tool breakage and chatter occurrence.
19 Mar 2018Go to complete version
AUTHORS: S. Garnier, K. Subrin, P. Arévalo-Siles, G. Caverot, B. Furet
Mobile Robot Stability for Complex Tasks in Naval Industries
Naval industries are looking for new ways to perform complex tasks on large and stationary parts. In this context, mobile manipulator which consists of a robot manipulator mounted on a wheeled base (Automated Guided Vehicle) is a promising solution. In order to guaranty the quality and free the workers to make dangerous or repetitive tasks, the manufacturing process is redefined where the worker supervises and collaborates with the robotic system. In this paper, the stability of such a robotic system is studied in a dynamic way for a better understanding of the stability of robotic system in order it to be a safe solution. Various representations are given to understand the influence of joint speed and joint acceleration. Finally, as a Key Performance Indicator, a map is drawn to represent the effect of the structure height on the tip-over.
10 Mar 2018Go to complete version
AUTHORS: A. Fabisch, C. Petzoldt, M. Otto, F. Kirchner
A Survey of Behavior Learning Applications in Robotics – State of the Art and Perspectives
Recent success of machine learning in many domains has been overwhelming, which often leads to false expectations regarding the capabilities of behavior learning in robotics. In this survey, we analyze the current state of machine learning for robotic behaviors. We will give a broad overview of behaviors that have been learned and used on real robots. Our focus is on kinematically or sensorially complex robots. That includes humanoid robots or parts of humanoid robots, for example, legged robots or robotic arms. We will classify presented behaviors according to various categories and we will draw conclusions about what can be learned and what should be learned. Furthermore, we will give an outlook on problems that are challenging today but might be solved by machine learning in the future and argue that classical robotics and other approaches from artificial intelligence.
5 Jun 2019Go to complete version
AUTHORS: W. Ji, L. Wang
Industrial Robotic Machining:
For the past three decades, robotic machining has attracted a large amount of research interest owning to the benefit of cost efficiency, high flexibility and multi-functionality of industrial robot. Covering articles published on the subjects of robotic machining in the past 30 years or so; this paper aims to provide an up-to-date review of robotic machining research works, a critical analysis of publications that publish the research works, and an understanding of the future directions in the field. The research works are organised into two operation categories, low material removal rate (MRR) and high MRR, according their machining properties, and the research topics are reviewed and highlighted separately. Then, a set of statistical analysis is carried out in terms of published years and countries. Towards an applicable robotic machining, the future trends and key research points are identified at the end of this paper.
3 Apr 2019Go to complete version
AUTHORS: W. Ji, Y. Li, L. Wang
A Task-Oriented Cyber-Physical System in Manufacturing
Towards automatic operations on a shop floor, this paper proposes a task-oriented cyber-physical system (CPS) concept, called holonic CPS. Within the context, operation processes are separated and modelled independently, and machines are also divided into hardware and basic controllers. The elementary distributed entities from operation processes and machines are represented by holons including “Cyber” and “Physical” parts. A holon is designed in hierarchical structure involving agent, FB, controller and hardware from top to bottom, and a holarchy, a holonic network including relevant holons, is able to represent a task or a function on the shop floor. To show how the proposed holonic CPS works, a robotic application is designed, in which both assembly tasks and machining (grinding and polishing) tasks can be performed. Where a set of relevant holons and holarchies are presented, which provide a “library” as the basis to represent an ordered task. A microkernel architecture is designed to implement the holonic CPS.
2 May 2019Go to complete version
AUTHORS: N. A.Theissen, T. Laspas, A. Archenti
Closed-force-loop Elastostatic Calibration of Serial Articulated Robots
This paper presents a novel methodology to measure the compliance of articulated serial robots based on the Elastically Linked Systems concept. The idea behind the methodology is to measure serial articulated robots with customized external wrench vectors under a closed-force-loop. The methodology proposes to measure robots in use-case defined configurations to increase the effect of the identified model parameters on their later implementation. The measurement methodology utilizes the Loaded Double Ball Bar to customize wrench vectors and a laser tracker to measure the system response. In particular, the Loaded Double Ball Bar creates the closed-force-loop to create a flow of forces similar to the intended application of the robot. The methodology is applied to an industrial robot with six rotary joints using the LDBB and a laser tracker. Finally, the paper ends on a discussion about the implementation of the model parameters to improve the accuracy of robots as well as challenges to realize a more cost efficient elastostatic calibration.
Jun 2019Go to complete version
AUTHORS: A. Fabisch
Empirical Evaluation of Contextual Policy Search with a Comparison-based Surrogate Model and Active Covariance Matrix Adaptation
Contextual policy search (CPS) is a class of multi-task reinforcement learning algorithms that is particularly useful for robotic applications. A recent state-of-the-art method is Contextual Covariance Matrix Adaptation Evolution Strategies (C-CMA-ES). It is based on the standard black-box optimization algorithm CMA-ES. There are two useful extensions of CMA-ES that we will transfer to C-CMA-ES and evaluate empirically: ACM-ES, which uses a comparison-based surrogate model, and aCMA-ES, which uses an active update of the covariance matrix. We will show that improvements with these methods can be impressive in terms of sample-efficiency, although this is not relevant any more for the robotic domain.
26 Oct 2018Go to complete version
AUTHORS: A. Fabisch
Pytransform3d is a Python library for transformations in three dimensions. Heterogenous software systems that consist of proprietary and open source software are often combined when we work with transformations. For example, suppose you want to transfer a trajectory demonstrated by a human to a robot. The human trajectory could be measured from an RGB-D camera, fused with inertial measurement units that are attached to the human, and then translated to joint angles by inverse kinematics. That involves at least three different software systems that might all use different conventions for transformations. Sometimes even one software uses more than one convention.
31 Jan 2019Go to complete version
AUTHORS: A. Fabisch
A Comparison of Policy Search in Joint Space and Cartesian Space for Refinement of Skills
Imitation learning is a way to teach robots skills that are demonstrated by humans. Transfering skills between these different kinematic structures seems to be straightforward in Cartesian space. Because of the correspondence problem, however, the result will most likely not be identical. This is why refinement is required, for example, by policy search. Policy search in Cartesian space is prone to reachability problems when using conventional inverse kinematic solvers. We propose a configurable approximate inverse kinematic solver and show that it can accelerate the refinement process considerably. We also compare empirically refinement in Cartesian space and refinement in joint space.
8 May 2018Go to complete version
AUTHORS: A. Verl, A. Valente, S. Melkote, C. Brecher, E. Ozturk, L. T. Tunc
Robotic machining centers offer diverse advantages: large operation reach with large reorientation capability, and a low cost, to name a few. Many challenges have slowed down the adoption or sometimes inhibited the use of robots for machining tasks. This paper deals with the current usage and status of robots in machining, as well as the necessary modelling and identification for enabling optimization, process planning and process control. Recent research addressing deburring, milling, incremental forming, polishing or thin wall machining is presented. We discuss various processes in which robots need to deal with significant process forces while fulfilling their machining task.
9 May 2019Go to complete version
AUTHORS: T. Muller, T. Bressac, S. Garnier, B. Furet, K. Subrin
Towards the Accuracy Improvement of a Mobile Robot for Large Parts Sanding
Recently, Naval industries explore ways to help workers with their activities. A mobile robot, including an automated guided vehicle and an industrial robot, is currently developed to address sanding of large parts. In this paper, we address the accuracy improvement of a mobile robot. The AGV, due to the industrial robot movement on his top, shows unwanted behavior and a positioning error of various millimeters due to the prismatic joints. Then, its behavior is identified with a theoretical approach and by measurement. Ways are then explored to validate the observation (camera and inertial system) to detect the unwanted movement. Perspectives are addressed to improve the accuracy of the mobile robot.
18-20 Mar 2019Go to complete version
AUTHORS: C. Sun, Patrick L.F. Kengne, A. Barrios, S. Mata, E. Ozturk
Form Error Prediction
in Robotic Assisted Milling
Robotic assisted milling is a process where a robot supports a workpiece while a machine tool cuts the workpiece. It can be used to suppress vibrations and minimize form errors in thin wall workpieces. In this paper, form error on a workpiece is simulated using a static force model, a frequency domain model and a hybrid model while a robot supports the workpiece from the other side. Machining results show that assistance of the robot has a considerable effect of the magnitude of form errors. Hence, support force should be carefully selected by simulation before machining. Finally, simulation results show that hybrid model gives the best fit among those three models.
Apr 2019Go to complete version
AUTHORS: L. Gutzeit, A. Fabisch, C. Petzoldt, H. Wiese, F. Kirchner
Automated Robot Skill Learning from Demonstration for Various Robot Systems
Transferring human movements to robotic systems is of high interest to equip the systems with new behaviors without expert knowledge. Typically, skills are often only learned for a very specific setup and a certain robot. We propose a modular framework to learn skills that is applicable on different robotic systems without adaptations. Our work builds on the recently introduced BesMan Learning Platform, which comprises the full workflow to transfer human demonstrations to a system, including automatized behavior segmentation, imitation learning, reinforcement learning for motion refinement, and methods to generalize to related tasks. For this paper, we extend this approach in order that different skills can be imitated by various systems in an automated fashion with a minimal amount of configuration, e.g., definition of the target system and environment. For this, we focus on the imitation of the demonstrated movements and show their transferability without movement refinement. We demonstrate the generality of the approach on a large dataset, consisting of about 700 throwing demonstrations.
24 Aug 2019Go to complete version
AUTHORS: M. Terreran, M. Antonello, S. Ghidoni
Boat Hunting with Semantic Segmentation for Flexible and Autonomous Manufacturing
Customized mass production of boats and other vehicles requires highly complex manufacturing processes that need a high amount of automation. To enhance the efficiency of such systems, sensing is of paramount importance to provide robots with detailed information about the working environment. In this paper, we propose the use of semantic segmentation to detect the key elements involved in production, to boost automation in the production process. Our main focus is on the sanding process of these tools by means of a robot. We demonstrate the potential of these techniques in an industrial environment featuring a lower degree of variability with respect to the domestic scenes typically considered in the literature. In the production environment, however, higher performances are required to address challenging manufacturing operations successfully. In this work, we also show that exploiting contextual cues and multiple points of view can further boost the reliability of our system, which provides useful data to the other robot modules in charge of navigation, work station recognition, and other tasks.
4-6 September 2019Go to complete version
AUTHORS: K. Subrin, S. Garnier, T. Bressac, B. Furet
Digital Chain Development for Sanding Application with a Kinematically Redundant Robotic System
Naval industries are looking for new ways to perform complex tasks on large and stationary parts. In this context, mobile manipulators which consist on a manipulator (6 Degrees of Freedom) mounted on a wheeled base (4 DoFs – Automated Guided Vehicule) are a promising solution. In this paper, such one is presented and the way to control it is explained. It appears that the redundancy management can be solved by fixing geometric parameters and adjusting timers to control the end effector. Staubli Robot and AGV are synchronised via a Main computer: the digital chain development is highlighted and ways are proposed to improve the mobile robot behavior via the redundancy management.
Apr 2019Go to complete version