Robotic Contact Tasks Research Articles

Robot Manipulation Skills Transfer for Sim-to-Real in Unstructured Environments

Robot force control that needs to be customized for the robot structure in unstructured environments with difficult-to-tune parameters guarantees robots’ compliance and safe human–robot interaction in an increasingly expanding work environment. Although reinforcement learning provides a new idea for the adaptive adjustment of these parameters, the policy often needs to be trained from scratch when used in new robotics, even in the same task. This paper proposes the episodic Natural Actor-Critic algorithm with action limits to improve robot admittance control and transfer motor skills between robots. The motion skills learned by simple simulated robots can be applied to complex real robots, reducing the difficulty of training and time consumption. The admittance control ensures the realizability and mobility of the robot’s compliance in all directions. At the same time, the reinforcement learning algorithm builds up the environment model and realizes the adaptive adjustment of the impedance parameters during the robot’s movement. In typical robot contact tasks, motor skills are trained in a robot with a simple structure in simulation and used for a robot with a complex structure in reality to perform the same task. The real robot’s performance in each task is similar to the simulated robot’s in the same environment, which verifies the method’s effectiveness.

Formal Verification of Robotic Contact Tasks via Reachability Analysis

Verifying the correct behavior of robots in contact tasks is challenging due to model uncertainties associated with contacts. Standard methods for testing often fall short since all (uncountable many) solutions cannot be obtained. Instead, we propose to formally and Efficiently verify robot behaviors in contact tasks using reachability analysis, which enables checking all the reachable states against user-provided specifications. To this end, we extend the state of the art in reachability analysis for hybrid (mixed discrete and continuous) dynamics subject to discrete-time input trajectories. In particular, we present a novel and scalable guard intersection approach to reliably compute the complex behavior caused by contacts. We model robots subject to contacts as hybrid automata in which crucial time delays are included. The usefulness of our approach is demonstrated by verifying safe human-robot interaction in the presence of constrained collisions, which was out of reach for existing methods.

Model-Based Coupling for Co-Simulation of Robotic Contact Tasks

Co-simulation of complex robotic systems allows the different components to be modeled and simulated independently using methods and tools tailored to their nature and time-scale, which makes the implementation process more modular and flexible. Some applications require the use of non-iterative coupling schemes for optimal performance, such as real-time interactive environments and human and hardware-in-the-loop setups. Stability of non-iterative schemes is challenging due to the restricted and delayed information that is exchanged between subsystems, and robust prediction of interface variables is key. Here, we propose a framework for exchanging model information between mechanical systems with contact, where reduced-order models approximate the interface dynamics of the subsystems. Effective mass and force terms are formulated using a reduced representation of the model, which can then be exchanged between subsystems and integrated in their simulation. The analysis of several simulations of challenging robotic contact tasks, such as grasping and insertion with jamming, shows that model-based coupling allows for stable co-simulation with larger interface stiffness values, resulting in stronger coupling and higher simulation accuracy.

Analysis of multimodal Bayesian nonparametric autoregressive hidden Markov models for process monitoring in robotic contact tasks

Robot introspection aids robots to understand what they do and how they do it. Previous robot introspection techniques have often used parametric hidden Markov models or supervised learning techniques, implying that the number of hidden states or classes is defined a priori and fixed through the entire modeling process. Fixed parameterizations limit the modeling power of a process to properly encode the data. Furthermore, first-order Markov models are limited in their ability to model complex data sequences that represent highly dynamic behaviors as they assume observations are conditionally independent given the state. In this work, we contribute a Bayesian nonparametric autoregressive Hidden Markov model for the monitoring of robot contact tasks, which are characterized by complex dynamical data that are hard to model. We used a nonparametric prior that endows our hidden Markov models with an unbounded number of hidden states for a given robot skill (or subtask). We use a hierarchical Dirichlet stochastic process prior to learn an hidden Markov model with a switching vector autoregressive observation model of wrench signatures and end-effector pose for the manipulation contact tasks. The proposed scheme monitors both nominal skill execution and anomalous behaviors. Two contact tasks are used to measure the effectiveness of our approach: (i) a traditional pick-and-place task composed of four skills and (ii) a cantilever snap assembly task (also composed of four skills). The modeling performance or our approach was compared with other methods, and classification accuracy measures were computed for skill and anomaly identification. The hierarchical Dirichlet stochastic process prior to learn an hidden Markov model with a switching vector autoregressive observation model was shown to have excellent process monitoring performance with higher identification rates and monitoring ability.

Normal and tangent force neuro-fuzzy control of a soft-tip robot with unknown kinematics

Assuming that contact kinematics is known, there exists many force robot control schemes, however the common practice of placing a deformable pad at contact makes difficult its implementation. The difficulty stems from the fact that the this pad introduces contact kinematics uncertainties due to the unknown deformation. In this paper, considering the full non-linear constrained rigid robot equipped with a hemispherical soft-tip as end-effector, a force regulator is proposed. To compensate the kinematic uncertainty at contact, induced by the unknown soft-tip deformation, a multi-input single-output (MISO) self-tuning fuzzy-rule emulated neural network (MiFRENN) is used. Additionally, the gravity compensation together with a damping injection term in the controller are used to guarantee local convergence of the normal and tangential force errors at a given equilibrium. The stability domain of the system varying depending on the knowledge-based contribution of the MISO-MiFRENN and the damping injection, which amounts for a novel scheme that can be used for other advanced robotic contact tasks, such as tactile exploration, dexterous manipulation or biped locomotion. Representative simulations illustrate the closed-loop numerical behavior.

The Robotic Force Tracking Based on Single Input Fuzzy Adaptive Controller

In order to realize precise contact tasks with an unknown environment, robotic force controllers have to adapt themselves to the unknown environment. In this paper, an effective controller for robot contact tasks is proposed using sliding mode switching surface as a sole fuzzy input variable for substituting the feedback controller in the model reference adaptive control (MRAC), the fuzzy force controller determines the adjustment to the position control loop, adjust the fuzzy controller parameters according to the difference between the actual and desired force responses. There is a sole fuzzy input variable in fuzzy logical controller(FLC), the total number of rules is greatly reduced compared to conventional FLC, hence generations and tuning of controller are more simple and easy. Simulations prove that the excellent adaptability as well as high precision is obtained by the proposed strategy.

Dynamics and Robust Control of Robot-environment Interaction

This book covers the most attractive problem in robot control, dealing with the direct interaction between a robot and a dynamic environment, including the human-robot physical interaction. It provides comprehensive theoretical and experimental coverage of interaction control problems, starting from the mathematical modeling of robots interacting with complex dynamic environments, and proceeding to various concepts for interaction control design and implementation algorithms at different control layers. Focusing on the learning principle, it also shows the application of new and advanced learning algorithms for robotic contact tasks. The ultimate aim is to strike a good balance between the necessary theoretical framework and theoretical aspects of interactive robots.

ON POST-IMPACT ANGULAR VELOCITIES AND RESULTANT IMPULSES WITH RANK-DEFICIENT JACOBIAN MATRICES USING NEWTON IMPACT LAW

Modeling and trustworthy simulation of impact play an important role in research on robotic contact tasks. Impact dynamic equations, based on Newton impact law, and their solution for planar multi-link robotic collisions have been well developed in literature in the context of determined contact problems. Rank-deficient Jacobian matrices cause the impact equations to be indeterminate. However this issue has not been investigated in previous research. In this paper, the solution for the velocity changes due to impact is proved to be unique in spite of rank-deficient Jacobian matrices and it is solved in a closed form that can be easily employed for simulating robotic system contact states. Furthermore, a set of linear equations with unknown impulses is obtained whereas the impulses can only be solved if extra contact constraints are provided. Two robot collision problems with rank-deficient Jacobian matrices are presented to exemplify the method.

Robot contact tasks in the presence of control target distortions

This work refers to the problem of controlling robot motion and force in frictional contacts under environmental errors and particularly orientation errors that distort the desired control targets and control subspaces. The proposed method uses online estimates of the surface normal (tangent) direction to dynamically modify the control target and control space decomposition. It is proved that these estimates converge to the actual value even though the elasticity and friction parameters are unknown. The proposed control solution is demonstrated through simulation examples in three-dimensional robot motion tasks contacting both planar and curved surfaces.

Impedance Learning for Robotic Contact Tasks Using Natural Actor-Critic Algorithm

Compared with their robotic counterparts, humans excel at various tasks by using their ability to adaptively modulate arm impedance parameters. This ability allows us to successfully perform contact tasks even in uncertain environments. This paper considers a learning strategy of motor skill for robotic contact tasks based on a human motor control theory and machine learning schemes. Our robot learning method employs impedance control based on the equilibrium point control theory and reinforcement learning to determine the impedance parameters for contact tasks. A recursive least-square filter-based episodic natural actor-critic algorithm is used to find the optimal impedance parameters. The effectiveness of the proposed method was tested through dynamic simulations of various contact tasks. The simulation results demonstrated that the proposed method optimizes the performance of the contact tasks in uncertain conditions of the environment.

Adaptive control of robot contact tasks with on-line learning of planar surfaces

In robot constrained motion problems on planar surfaces with frictional contacts, uncertainties on the contacted surface not only affect the control system performance but also distort control targets. The surface normal direction cosines are in this case uncertain parameters that are involved in both the control law and the control targets. This work proposes an adaptive learning controller that uses force and joint position/velocity measurements to simultaneously learn the surface orientation and achieve the desired goal. Simulation examples for a 6 dof robot are used to illustrate the theoretical results and the performance of the proposed controller in practical cases.

Blind force/position control on unknown planar surfaces

In robot contact tasks on planar surfaces with frictional contacts, kinematic uncertainties of the contacted surface affect the control system performance and distort control targets. An adaptive controller that uses force and joint position/velocity measurements to simultaneously learn the surface orientation and achieve the desired goal is proposed. Simulation examples are used to illustrate the theoretical results and the performance of the proposed controller in practical cases.

Updating workpiece geometry using robotic sensor information gathered during contact tasks

During robotic contact tasks, geometric information of the workpiece is used to specify the position of the robot’s hand on the workpiece and the direction of force control. This geometry is idealized in a typical CAD file, but due to manufacturing precision or wear, the actual workpiece geometry is inevitably deviated from the desired geometry. Furthermore, when the workpiece is mounted, position and orientation inaccuracies emerge. In this paper, we investigate two questions: (1) Can the workpiece geometry in the CAD file be used to control a robot in contact with an inaccurately placed workpiece?; and (2) Once the task is performed, how can the robot’s sensor information be used to update the geometry of the workpiece? A methodology is developed to solve robotic control problems with workpiece position and geometry inaccuracies. Once performed, the CAD file image is displaced to fit the sensed trajectory of the robot’s hand. Finally, the workpiece image geometry is modified using a least squares approximation to fit the sensed data more accurately. In the end, the robot performs the contact task while gathering information that is used to update the original CAD file geometry. The methodologies are demonstrated through a simulation experiment that requires a robot to shave a geometrically altered face that is inaccurately positioned.

Adaptive Modular Vector Field Control for Robot Contact Tasks in Uncertain Environments

We propose adaptive modular vector field control (AMVFC) for a robot manipulator to interact with uncertain environmental geometric constraints. Starting from an uncertain geometric model of the environment, we first parameterize the desired velocity vector field of the robot using the weighted combination of a set of basis vector fields. Then, to overcome the influence of environmental model uncertainty, we add force feedback to adjust robot dynamics and the weight parameters of the desired velocity field for the robot to approach the real environment. Simulation of a robot interacting with uncertain circles and an ellipse demonstrates the effectiveness of our approach.

ロボットによる接触作業における内力の解析

Robots may suffer from excessive internal force in contact tasks especially when they are position-controlled or executing complicated assemblies. Excessive internal force will lead to failure of the tasks, so it is important to predict and avoid such situations. This paper presents mechanical analysis of internal force applied to objects in robotic manipulation. A linear-programming-based algorithm that can judge the possibility of excessive internal force under Coulomb friction is proposed. Numerical examples including an internal force problem in graspless manipulation and a wedging problem in peg-in-hole are also showed. Our algorithm will help the planning of robotic contact tasks.

Control of contact via tactile sensing

We present our approach to using tactile sensing in the feedback control of robot contact tasks. A general framework, called tactile servo, is first introduced. We then address the critical issue of how to model the state of contact in terms that are both sufficient for defining general contacts and conducive to bridging the gap between a robot task description and the information observable by a tactile sensor. We subsequently examine techniques for deriving tactile sensor models that are required for computing the state of contact from the sensor output. Two basic methods for mapping a tactile sensor image to the contact state variables are introduced: one based on explicit inverse modeling, and the other on numerical tactile Jacobian. In both cases, moments of the tactile sensor image are used as the features that capture the necessary information about contact. The theoretical development is supported by extensive experiments, which include edge tracking, object shape construction and object manipulation.

Robot Manipulator Force Control Using Fuzzy-Neuro. Fuzzy Selection of Fuzzy-Neuro Controllers According to Environment Properties.

Force control is one of the most important and fundamental tasks of robot manipulators. In order to realize precise contact tasks with an nuknown environment, robot controllers have to adapt themselves to the unknown environment. Recently, fuzzy reasoning and/or neural networks are expected to play an important role in force control of robot manipulator. Human knowledge can be reflected in control rules by using fuzzy control method, and learning/adaptation ability can be obtained by applying neural networks to robot controllers. In this paper, an effective control policy for robot contact tasks with an unknown environment is proposed using fuzzy-neuro techniques. In this control policy, a neural network is applied to classify the unknown environment based on its dynamic response of the environment and then select the suitable fuzzy neural force controller. The selected fuzzy neural force controller is able to realize the desired contact force precisely using its on-line adaptation ability. Furthermore, the fuzzy selection of the controller is introduced for force control with the environment whose suitable fuzzy neural force controller is not prepared. The effectiveness of the proposed method is evaluated by experiment with a 2DOF planar robot manipulator.

A Natural Framework for Designing Bounce-Less Controller for Robotic Contact Tasks

Abstract In robotic tasks where the manipulator has to make transition from free space motion to constrained one, there always exists a inevitable phase transition. A number of controllers have been proposed in the literature with various discussions on their practical implications. In this paper for the first time a novel framework for studying design of controllers for robotic contact tasks problem is proposed. This framework presents a natural set-up to study the performance of controller. The method of this paper combines the embedded passive compliant properties of the interacting system (i.e. manipulator and the contacting environment) with the basic controller to achieve the hybrid bounce-less property. Here the interacting environment is used as a natural switching surface for the discontinuous controller. The application of the proposed framework is demonstrated through experimental results.

The Application of Connectionist Structures to LearningImpedance Control in Robotic Contact Tasks

The goal of this paper is to consider the synthesis of learning impedance control using recurrent connectionist structures for on-line learning of robot dynamic uncertainties in the case of robot contact tasks. The connectionist structures are integrated in non-learning impedance control laws that are intended to improve the transient dynamic response immediately after the contact. The recurrent neural network as a part of hybrid learning control algorithms uses fast learning rules and available sensor information in order to improve the robotic performance progressively for a minimum possible number of learning epochs. Some simulation results of deburring process with the MANUTEC r3 robot are presented here in order to verify the effectiveness of the proposed control learning algorithms.

Stabilizing position/force control of robots interacting with environment by learning connectionist structures

This paper is mostly concerned with the application of connectionist architectures for fast on-line learning of robot dynamic uncertainties used at the executive hierarchical control level in robot contact tasks. The connectionist structures are integrated in non-learning control laws for contact tasks which enable stabilization and good tracking performance of position and force. It has been shown that the problem of tracking a specified reference trajectory and specified force profile with a present quality of their transient response can be efficiently solved by means of the application of a four-layer perceptron. A four-layer perceptron is part of a hybrid learning control algorithm through the process of synchronous training which uses fast learning rules and available sensor information in order to improve robotic performance progressively in the minimum possible number of learning epochs. Some simulation results of the deburring process with robot MANUTEC r3 are shown to verify effectiveness of the proposed control learning algorithms.