Iterative Learning Control for Compliant Underactuated Arms

The dynamic manipulation of deformable objects poses a significant challenge in robotics. While model-based approaches for controlling such objects hold significant potential, their effectiveness hinges on the availability of an accurate and computationally efficient dynamics model. This work focuses on sample-efficient learning of models to capture the dynamic behavior of deformable linear objects (DLOs). Inspired by the pseudo-rigid body method, we present a physics-informed neural ODE that approximates a DLO as a serial chain of rigid bodies interconnected by passive elastic joints. However, unlike traditional uniform spatial discretization and linear springdamper joints, our approach involves learning-based discretization and nonlinear elastic joints that characterize interaction forces via a neural network. Through real-world and simulation experiments involving DLOs with markedly different physical properties, we demonstrate the model's ability to accurately predict DLO motion.

This paper shows a spatial model of a closed elastica under external forces. The model is a discrete elastica model approximated by a serial chain of rigid links connected with passive elastic joints. The effect of external forces on a closed elastica is considered based on knowledge on mechanics of the multi-link system. We also show that it is possible to find the dimensionless form of the discrete elastica model successfully.

Fabricating a soft robot using conventional molding methods is difficult and time-consuming. Moreover, the types of materials used in the process are limited, and the elasticity cannot be changed incrementally. In this paper, we explain the detailed process of manufacturing molds for silicone joints. We construct a prototype molded silicone joint. We measure the elastic modulus of this joint and confirm that the elastic modulus and anisotropy change depending on the density, size, and arrangement of the surface grooves in the mold. We also develop a prototype human phantom using the proposed joint. We aim to contribute to the medical field by applying new techniques made possible by soft robotics.

We have confirmed that passive elastic trunk joints affect the energy efficiency of quadruped robots. The main causes of energy loss are negative work and contact between the legs and ground. We clarified relationship between legs and trunk joints for decreasing energy loss. Elastic energy of the trunk joints increases at the timing when kinetic energy in robots decreases. On the other hand, elastic energy of the trunk joints is released at the timing when kinetic energy in robots increases. Energy loss is smaller by this mechanism. In this paper, we develop a quadruped walking robot having passive elastic joints in the trunk. Also, by this walking robot, we verify that the proposed relationship between leg and trunk joints is reliable.

A closed kinematic chain, like an arm that operates a crank, has a constrained movement space. A meaningful movement of the chain's endpoint is only possible along the free movement directions which are given implicitly by the contour of the object that confines the movement of the chain. Many technical solutions for such a movement task, in particular those used in robotics, use central controllers and force-torque sensors in the arm's wrist or a leg's ankle to construct a coordinate system (task frame formalism) at the local point of contact the axes of which coincide with the free and constrained movement directions. Motivated by examples from biology, we introduce a new control system that solves a constrained movement task. The control system is inspired by the control architecture that is found in stick insects like Carausius morosus. It consists of decentral joint controllers that work on elastic joints (compliant manipulator). The decentral controllers are based on local positive velocity feedback (LPVF). It has been shown earlier that LPVF enables contour following of a limb in a compliant motion task without a central controller. In this paper we extend LPVF in such a way that it is even able to follow a contour if a considerable counter force drags the limb away along the contour in a direction opposite to the desired. The controller extension is based on the measurement of the local mechanical power generated in the elastic joint and is called power-controlled relaxation LPVF. The new control approach has the following advantages. First, it still uses local joint controllers without knowledge of the kinematics. Second, it does not need a force or torque measurement at the end of the limb. In this paper we test power-controlled relaxation LPVF on a crank turning task in which a weight has to be winched up by a two-joint compliant manipulator.

Deals with the modeling of a flexible inverted pendulum. The mathematical model of the controlled object used is a simplified one, which is approximated by rigid links and springs. The possibility of driving a simplified model of the flexible inverted pendulum using a double inverted pendulum with an elastic joint is studied. The experimental results show that the simplified model is sufficient to describe the flexible inverted pendulum, but with some modeling and identification errors. These uncertainty will excite the vibration. An H/sub infinity / control to stabilize the flexible inverted pendulum using the simplified model is presented. >

The repetitive processing and large quantity of single product represented by 3C products are urgently needed. However, for current processing operations, previous processing data have not been used in the optimization of control input. In order to utilize previous processing data to facilitate the next process and avoid adverse effects caused by repetitive disturbance and noise, the idea of iterative learning was introduced to improve the accuracy of machining. On the control level, since it is difficult to obtain high accuracy by traditional feedback control when faced with complex trajectories, an open-loop iterative learning controller and a position loop feedback controller were introduced, which worked fast with good convergence effects. Aiming at reducing the influence of accidental error, step type iterative learning was put forward. The iteration mechanism was stopped when the accuracy converged to the allowable range so as to reduce computational complexity, store the current iterative part of the control input, and make constant value compensation. However, in simulation and experiment, it was found that after superposition of the iterative learning controller, the phenomenon of partial divergence of the system tracking error occurred. Therefore, the speed and acceleration characteristics of input trajectories in time domain and frequency domain were analyzed. High-frequency noise was introduced in frequency domain, which was found to be the cause of the abovementioned phenomenon, and high-frequency components were filtered to solve the problem. To further improve the accuracy of convergence and avoid filtering effective high-frequency information in some area, a switchable filter based on the analysis of the frequency characteristics of input trajectory was proposed. Through SIMULINK simulation and dSPACE experimental verification, it was proved that the iterative learning controller of modifying controlled quantity and filter based iterative learning control method are effective.

Position control of an underactuated manipulator that has one passive joint is investigated. The dynamic constraint caused by the passive joint is second-order nonholonomic. Time-scaling of the active joint trajectory and bi-directional motion planning from the initial and the desired configurations provide an exact solution of the positioning trajectory. The active and passive joints can be positioned to the desired angles simultaneously by swinging the active joints only twice. Feedback control constrains the manipulator along the planned path in the configuration space. Simulation and experimental results show the validity of the proposed methods.

Position control of an underactuated manipulator that has one passive joint is investigated. The dynamic constraint caused by the passive joint is second-order nonholonomic. Time-scaling of the active joint trajectory and bi-directional motion planning from the initial and the desired configurations provide an exact solution of the positioning trajectory. The active and passive joints can be positioned to the desired angles simultaneously by swinging the active joints only twice. Feedback control constrains the manipulator along the planned path in the configuration space. Simulation and experimental results show the validity of the proposed methods.

A parallel mechanism’s pose is usually obtained indirectly from the active joints’ coordinates by solving the direct kinematics problem. Its accuracy mainly depends on the accuracy of the measured active joints’ coordinates, the tolerances in the active and passive joints, possible backlash, axes misalignment, limb deformations due to stress or temperature, the initial pose estimate that is used for the numerical method, and the accuracy of the kinematic model itself. Backlash and temperature deformations in the active joints especially hinder high-precision applications as they usually cannot be observed. By implementing a camera module on the base platform and an array of fiducial tags on the moveable manipulator platform of a parallel mechanism, a highly accurate, direct, and absolute pose measurement system can be obtained that can overcome those limitations. In this paper, such a measurement system is proposed, designed, and its accuracy is investigated on a state-of-the-art H-811.I2 6-axis miniature hexapod by Physik Instrumente (PI) GmbH & Co. KG.

The classical problem setup of iterative learning control (ILC) is to enforce tracking of a reference profile specified at all time points in the fixed task duration. The removal of the time specification releases significant design freedom in how the path is followed but has not been fully exploited in the literature. This article unlocks this extra design freedom by formulating the ILC task description to handle repeated path-following tasks, e.g., welding and laser cutting, which aim at following a given “spatial” path defined in the output space without any temporal information. The general ILC problem is reformulated for ILC design with the inclusion of an additional performance index, and the class of piecewise linear paths is characterized for the reformulated problem setup. A two-stage design framework is proposed to solve the characterized problem and yields a comprehensive algorithm based on an ILC update and a gradient projection update. This algorithm is verified on a gantry robot experimental platform to demonstrate its practical efficacy and robustness against model uncertainty.

<p>Traditional aerial manipulation systems were usually composed of rigid-link manipulators attached to an aerial platform, arising several rigidity-related issues such as difficulties of reach, compliant motion, adaptability to object’s shape and pose uncertainties, and safety of human-manipulator interactions, especially in unstructured and confined environments. To address these issues, partially compliant manipulators, composed of rigid links and compliant/flexible joints, were proposed; however, they still suffer from insufficient dexterity and maneuverability. In this article, a new set of compliant aerial manipulators is suggested. For this purpose, the concept of aerial continuum manipulation system (ACMS) is introduced, several conceptual configurations are proposed, and the functionalities of ACMSs for different applications are discussed. Then, the performances of proposed aerial manipulators are compared with conventional aerial manipulators by implementing available benchmarks in the literature. To enhance the comparison, new features with related benchmarks are presented and used for evaluation purposes. In this study, the advantages of ACMSs over their rigid-link counterparts are illustrated and the potential applications of ACMSs are suggested. The open problems such as those related to dynamic coupling and control of ACMSs are also highlighted.</p>

Ionomeric polymer-metal composites (IPMC) are electroactive polymer (EAP) materials that bend when electrically stimulated. As IPMC is a relatively new material, proper control methods have not yet evolved. In this paper the usability of IMPC actuators in real world applications is examined from the point of view of precision control. We propose a classical inverted pendulum control problem as a testbed. We suggest that if the pendulum can be balanced, then we have proven the usability of IPMC actuators for precise control tasks. In this paper we describe an inverted pendulum system driven by an IPMC actuator with PC-based control and a camera in the feedback loop. We report the preliminary experimental results in controlling the pendulum and discuss further improvements. To our knowledge this is the first attempt to control a system or manipulate an object with IMPC actuators in a feedback loop

A method is proposed for controlling the position of a manipulator with passive joints that have holding brakes instead of actuators. In this method, the coupling characteristics of manipulator dynamics are used, and no additional mechanisms are required. The effectiveness of the method was verified by experiments using a prototype manipulator. The prototype is a two-degree-of-freedom, horizontally articulated manipulator. The first axis is an active joint, and the second axis is a passive joint. While the brake of the passive joint is released, the passive joint is indirectly controlled by the motion of the active joint through the use of dynamic coupling. While the brake is engaged, the active joint is controlled. By combining these two control modes, the total position of the manipulator is controlled. The experiments show that use of this method makes the precise positioning of the passive joints possible.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

An Iterative Learning Control Design for Self-Servowriting in Hard Disk Drives