Many optimization problems in robotics involve the optimization of time-expensive black-box functions, such as those involving complex simulations or evaluation of real-world experiments. Furthermore, these functions are often stochastic as repeated experiments are subject to unmeasurable disturbances. Bayesian optimization can be used to optimize such methods in an efficient manner by deploying a probabilistic function estimator to estimate with a given confidence so that regions of the search space can be pruned away. Consequently, the success of the Bayesian optimization depends on the function estimator’s ability to provide informative confidence bounds. Existing function estimators require many function evaluations to infer the underlying confidence or depend on modeling of the disturbances. In this paper, it is shown that the confidence bounds provided by the Wilson Score Kernel Density Estimator (WS-KDE) are applicable as excellent bounds to any stochastic function with an output confined to the closed interval [0;l] regardless of the distribution of the output. This finding opens up the use of WS-KDE for stable global optimization on a wider range of cost functions. The properties of WS-KDE in the context of Bayesian optimization are demonstrated in simulation and applied to the problem of automated trap design for vibrational part feeders.

In this paper, a new method for offline programming part assembly operations with tight fittings is presented. More specifically, an assembly process trajectory generator with self programming capabilities is developed where the user needs to provide only very sparse and intuitive input. The presented system is added to the existing skill based robot software package VEROSIM. In VEROSIM, the trajectory generator is applied to an industrial test platform for assembling insulin injection devices at the Danish pharmaceutical company Novo Nordisk, where it is shown that the trajectories as expected are executable. Hence, the method is a strong alternative to online approaches such as programming by demonstration.

In this paper, we present a model based approach for predicting the forward kinematics of a tool for picking prepreg fiber plies from a flat table and draping them onto a double curved mold. The tool consists of suction cups interlinked with springs. The tool has 60 actuated and 240 passive degrees of freedom. The prediction is based on establishing a kinematic model of the total potential energy of the tool and minimizing this energy, under the assumptions that dynamic artifacts can be neglected as the movements are relatively slow, and that the springs interpolate the orientation of the suction cups. Various assumptions and simplifications are presented to increase computational speed.Finally, the model predictions are compared to real configurations measured by a camera. Four configurations are simulated and measured and the largest RMS positional deviation measured is 1.99mm, the largest residual measured is 3.68mm. The tolerance of the layup process is 2.5mm so more work needs to be done to reduce the maximum deviations.

Execution of automatically generated programs for accurate robotic machining requires the generated trajectories to be not only accurate with respect to the work piece, but also that the trajectories are continuous differentiable (C1) while avoiding unnecessary large curvatures leading to large accelerations that could compromise machining quality or speed. A widely used work piece representation is 3D triangle meshes as they can be easily generated in any CAD representations and from surface scans, and they are also very suitable for robotics applications. However, they lack the C1property across the triangle edges.In this paper, a new method for generating C1surfaces based on 3D triangle meshes is presented. It will be shown by an example that the method is as good as existing methods with respect to the accuracy of the generated surface, and that the problem with large curvatures is much smaller than for existing methods. Moreover, the difficult input specification of derivatives at the vertices is avoided with this method.

We propose a new function estimator, called Wilson Score Kernel Density Estimation, that allows to esti-mate a mean probability and the surrounding confidence interval for parameterized processes with binomiallydistributed outcomes. Our estimator combines the advantages of kernel smoothing, from Kernel Density Esti-mation, and robustness to low number of samples, from Wilson Score. This allows for more robust and dataefficient estimates compared to the individual use of these two estimators. While our estimator is generallyapplicable for processes with binomially distributed outcomes, we will present it in the context of iterativeoptimization. Here we first show the advantage of our estimator on a mathematically well defined problem,and then apply our estimator to an industrial automation process.

In this paper we present an initial approach towards reversible robot movement primitives. Our approach is a modification of Dynamic Movement Primitives (DMPs), a widely used framework for robot learning from demonstration. DMPs are based on dynamical systems to guarantee properties such as convergence to a goal state, robustness to perturbation, and the ability to generalize to other goal states. Yet a main limitation of their original formulation is that they do not allow for movements to be reversed. Thus, to execute the same task forwards and backwards would mean to learn two separate primitives. We propose to replace the transformation system in DMPs with the Logistic Differential Equation (LDE), a known time-reversible non-linear system. Similarly to the original DMP formulation, our system's temporal evolution is controlled by a phase system, which in our case is derived from the LDE to guarantee reversibility. We evaluate our approach experimentally with demonstration data from a real robot assembly task, and show comparable properties to those of the original DMP system.

This paper presents a method for numerically computing the highest path velocity that a collaborative robot can attain, while complying with safety requirements. The safety requirements are obtained from ISO/TS 15066 that describes a collaborative method called power and force limiting, which specifies safe collisions between humans and robots. In particular, we assume that a path is given and compute the point-wise maximal path velocity that ensures a safe impact, i.e., the paper provides no considerations on the post impact safety.

Automated assembly operations have traditionally been carried out using “hard automation”, where all parts to be handled during the operation are located at known positions with very high accuracy. Such automation solutions are often mechanically very expensive to implement and also the programming time can be substantial. Hence, hard automation is only suitable for large batch sizes. However, with the upcoming programming paradigms based on joining preprogrammed subtasks (skills) to a complete solution, the programming time can be shortened significantly. Skills are subtasks with well defined interfaces in terms of pre- and postconditions. However, skills still rely on high accuracy as there is no methodology for handling the propagation of inaccuracies when combining the skills. Therefore, skills are typically only useful for hard automation situations. In cases with smaller batch sizes, the high accuracy property will have to be loosened to lower the costs on the hardware side. In this paper, we extend the skill interfaces with a unified formalism for how the skill propagates inaccuracies. The formalism includes how the inaccuracy propagation for a skill can be learned and how the learned propagations can be used to join the skills while also predicting how inaccuracies will propagate through the entire skill chain. We also briefly discuss the perspectives using the formalism to offline search for the most robust combinations of skills for the entire task.

In this paper, we suggest a coherent way of representing results from experiments associated with robotic assembly. The purpose of the representation is to be able to reuse the experiments in other assembly settings. A main novelty in our representation is the inclusion of fine grained experimental uncertainties such as e.g. deviations between a sensed object pose and the actual pose, and we discuss why it is very important for the reusability of experiments to include these uncertainties. Under the reasonable assumption that we can represent the uncertainties as a region around the origin in a potentially high dimensional Cartesian space, we show that we can efficiently represent the studied deviations by storing experiments on a so called spherical lattice. We illustrate that the representation works by studying simulation experiments on two different industrial use cases involving grasping an object and mounting an object on a fixture.

Learning action parameters is becoming an ever more important topic in industrial assembly with tendencies towards smaller batch sizes, more required flexibility and process uncertainties. This paper presents a statistical online learning method capable of handling these issues. The method uses elimination of unpromising parameter sets to reduce the elements of the discretised sample space (inspired by Action Elimination) based on regression uncertainty. Kernel Density Estimation and Wilson Score are explored as internal representations. Based on a dynamic simulator setup for a real world Peg-in-Hole problem, it is shown that the presented method can drastically reduce the number of samples needed. Furthermore, it is also shown that the solution obtained in simulation by our learning method succeeds when executed on the corresponding real world setup.

Simulation in robotics is often based on physics engines developed for computer games and animation. These engines are inaccurate due to the applied first order numerical integration and because their ways to resolve redundant contacts lead to contact forces that are distributed in a physically unreasonable way among multiple contacts. In this paper, we present a new physics engine that resolves these flaws by designing and implementing a second order integration method, and by a new way of defining the contact-force resolution problem. We illustrate the improvement over state-of-the-art engines by a number of experiments. The new engine is proven to be particularly suitable for simulating tight fitting assembly operations where multiple contacts occur between the objects involved.

We present a method for finding correspondence between 3D models. From an initial set of feature correspondences, our method uses a fast voting scheme to separate the inliers from the outliers. The novelty of our method lies in the use of a combination of local and global constraints to determine if a vote should be cast. On a local scale, we use simple, low-level geometric invariants. On a global scale, we apply covariant constraints for finding compatible correspondences. We guide the sampling for collecting voters by downward dependencies on previous voting stages. All of this together results in an accurate matching procedure. We evaluate our algorithm by controlled and comparative testing on different datasets, giving superior performance compared to state of the art methods. In a final experiment, we apply our method for 3D object detection, showing potential use of our method within higher-level vision.

This paper describes an adaptable system which is able to perform manipulation operations (such as Peg-in-Hole or Laying-Down actions) with flexible objects. As such objects easily change their shape significantly during the execution of an action, traditional strategies, e.g, for solve path-planning problems, are often not applicable. It is therefore required to integrate visual tracking and shape reconstruction with a physical modeling of the materials and their deformations as well as action learning techniques. All these different submodules have been integrated into a demonstration platform, operating in real-time. Simulations have been used to bootstrap the learning of optimal actions, which are subsequently improved through real-world executions. To achieve reproducible results, we demonstrate this for casted silicone test objects of regular shape. Note to Practitioners - The aim of this work was to facilitate the setup of robot-based automation of delicate handling of flexible objects consisting of a uniform material. As examples, we have considered how to optimally maneuver flexible objects through a hole without colliding and how to place flexible objects on a flat surface with minimal introduction of internal stresses in the object. Given the material properties of the object, we have demonstrated in these two applications how the system can be programmed with minimal requirements of human intervention. Rather than being an integrated system with the drawbacks in terms of lacking flexibility, our system should be viewed as a library of new technologies that have been proven to work in close to industrial conditions. As a rather basic, but necessary part, we provide a technology for determining the shape of the object when passing on, e.g., a conveyor belt prior to being handled. The main technologies applicable for the manipulated objects are: A method for real-time tracking of the flexible objects during manipulation, a method for model-based offline prediction of the static deformation of grasped, flexible objects and, finally, a method for optimizing specific tasks based on both simulated and real-world executions.

We address the problem of estimating the alignment pose between two models using structure-specific local descriptors. Our descriptors are generated using a combination of 2D image data and 3D contextual shape data, resulting in a set of semi-local descriptors containing rich appearance and shape information for both edge and texture structures. This is achieved by defining feature space relations which describe the neighborhood of a descriptor. By quantitative evaluations, we show that our descriptors provide high discriminative power compared to state of the art approaches. In addition, we show how to utilize this for the estimation of the alignment pose between two point sets. We present experiments both in controlled and real-life scenarios to validate our approach.

Automatic offline programming of industrial robotic systems is becoming increasingly important due to the larger percentage of desired automation of low volume tasks. Often, such tasks may involve handling of items that can have rather large deflections which are important to take into account when doing offline programming. In this paper such a problem is presented, namely robotic assembly of unique concrete reinforcement structures. Reinforcement bars of 3 meters may deflect up to around 50cm. We illustrate experimentally how the reinforcement bar can be precisely modelled by a structure consisting of rigid parts connected by “deflection joints”. Such a model can be directly integrated into existing physics simulation engines such as the Open Dynamics Engine (ODE). Finally, we discuss how the simulation will be used for automatic offline programming and present a video with a dynamic simulation of the reinforcement assembly process.

In this paper, we present what appears to be the first studies of how to apply learning methods for improving the grasp success probability in industrial bin picking. Our study comprises experiments with both a pneumatic parallel gripper and a suction cup. The baseline is a prioritized list of grasps that have been chosen manually by an experienced engineer. We discuss generally the probability space for success probability in bin picking and we provide suggestions for robust success probability estimates for difference sizes of experimental sets. By performing grasps equivalent to one or two days in production, we show that the success probabilities can be significantly improved by the proposed learning procedure.

To understand the capabilities and behavior of a robot it is important to have knowledge about its physical structure and how its actuators control its shape. In this paper we analyze the kinematics and develop a general representation of a configuration of the heterogeneous modular robot Odin. The basics of estimating the shape of the Odin robot is presented, which leads the way for further research on the Odin robot and similar robots. We present an example of how to represent and estimate the shape of a tetrahedron configuration with various types of modules. We conclude that this representation can be used to find the physical constraints of the Odin robot and estimate the shape of a configuration.

We present a roadmap based planner for finding robot motions for gantry mounted manipulators for a line welding application at Odense steel shipyard (OSS). The robot motions are planned subject to constraints on when the gantry may be moved. We show that random sampling of gantry configurations is a viable technique for positioning the manipulator and present a pruning technique for managing the growth of the roadmap. We discuss results from simulations and from applications at the shipyard, where a similar planner has now been implemented for production.

In this paper we present a new concept for gripping objects in industrial applications. We assume that a priori, the object pose is only known with a relative low accuracy. Despite this, our method can lead to high accuracy gripping suitable for e.g. industrial assembly. Our concept is to augment a simple parallel gripper by mounting a set of object specific jaws. Given the right shapes these jaws enable the gripper to automatically align the object, and thereby compensate for errors in the original object pose estimation. We introduce a couple of automatic and semi-automatic design strategies for deriving these shapes and describe how these can be verified by simulations using rigid body dynamics. The output of our system is a CAD model for the jaws. We prove our concept with a number of experiments, which are also used to verify the coherence between the simulation and real world

Automatic specification of desired motions and planning of optimal feature sets are fundamental problems which must be addressed in order to ensure accurate and robust execution of complex tasks using visual servoing. In this paper, we introduce a simulation platform for generation of visual servoing tasks as well as a new method for automatic selection of object features. This method considers several aspects related to the robustness of the tracking system to select features which ensure optimal performance of the robot control system. Finally, we present experiments which compare the performance of a visual servoing system employing the proposed feature planning technique to that of a servoing system based on features selected using traditional methods. These experiments demonstrate that our technique not only improves the robustness of the visual tracking system but also significantly increases the accuracy of the visual servoing control loop.

The need for open model-based robot controllers becomes more and more important with the increasing quality of autonomous joint motion planners and the increasing amount of applications for these. An important item in a model-based controller is a precise dynamical model of the robot. Such a model is designed in terms of various inertial and friction parameters that must be either measured directly or determined experimentally. In this paper, we present a new method for determining parameters of a dynamical robot model based on experiments. We compare the method with existing methods both theoretically and experimentally, and show that we achieved some improvements in the parameter values by using the new method. The experiments were carried out on the two first links of a seven-axis Mitsubishi PA-10 robot.

This paper reports the first phase of a project whose aim is the automatic generation of tool center trajectories for robots engaged in spray painting of arbitrary surfaces. The first phase consists of proposing a mathematical model for the paint flux field within the spray cone. We have called this quantity the paint flux field partly to emphasize that it is a vector field and partly to distinguish it from a paint flux distribution function, which describes the angular variation of the flux field within the spray cone. It is shown that this flux field can be derived from experimental measurements performed by robots, of coverage profiles of paint strips on flat plates by solving a singular integral equation. This flux field is derived both for published experimental data as well as two sets of data from experiments performed by the authors. The correctness of the model is demonstrated by using the underlying distribution function to predict coverage profiles for other experiments in which the spray gun is no longer vertical to the plane surface.