Jean-Claude Piedboeuf

Task verification facility for the Canadian special purpose dextrous manipulator

As a partner in the International Space Station, Canada is responsible for the verification of all tasks involving the special purpose dextrous manipulator (SPDM). In this paper, the concept of an SPDM task verification facility (STVF) is described. The verification process involves three complementary stages. First, a real-time software simulator is used to verify the complete nominal and malfunction procedures. Next, the feasibility of a task involving contact is verified with the help of a hardware-in-the-loop simulator. Finally, a non real-time simulator is used to perform detailed parametric studies given the known tolerances on the components. The paper describes all three stages but emphasizes on the hardware-in-the-loop simulator, the only new facility within the STVF.

Dynamics modeling and simulation of constrained robotic systems

Dynamic analysis is the basic element of mechanical design and control of mechanisms. This work intends to address dynamic methods relevant to constrained robots and mechanisms from a unified analytical point of view, which is based on differential variational principles. A constrained robotic system is a mechanical system, where we need to consider kinematic constraint conditions explicitly in dynamic modeling and analysis. Important classes of constrained robotic systems include, for example, parallel robots and closed-chain mechanisms where the loop closure conditions can be generally expressed by nonlinear holonomic constraint equations, and mobile robots where the system is subjected to linear nonholonomic constraints. Our primary focus is on systems with nonlinear holonomic constraint equations (e.g., parallel robots, robotic systems with closed kinematic chains). However, the approach and formulation discussed are also applicable for nonholonomic systems. In the framework presented, many approaches can be discussed, and new directions can be highlighted that can contribute to the better understanding of dynamic behavior. Two new approaches for the dynamic analysis and for the simulation of constrained robotic systems are introduced and discussed. The paper also points out some areas and methods where further exploration is necessary to shed light on problems and applications related to constrained robotic systems.

Contact dynamics emulation for hardware-in-loop simulation of robots interacting with environment

To verify all robotic tasks involving a space robot interacting with environment, such as the special purpose dexterous manipulator, one should appeal to a simulation technique because the space robot cannot operate in an 1-g environment. However, to simulate dynamical behavior of a robot interacting with environment possess many difficulties due to complexity of the physical phenomenon involved during the interaction. In this work we develop an hardware-in-loop simulation technique, where a simulation of the space robot dynamics is combined with emulation of the contact dynamics by using a rigid robot prototype performing the contact task. The rigid robot is not dynamically or kinematically equivalent to the space robot, but it is controlled so that its endpoint dynamics replicates that of the space robot. Experimental results given from implementation of a single axis arm are presented.

Emulation of a space robot using a hydraulic manipulator on ground

To verify the tasks of one of the Space Station manipulator, the Canadian Space Agency is developing a hardware-in-the-loop simulation system. A hydraulic robot manipulating mockup of the space payload reproduces the contact forces occurring during insertion and extraction. The control law allowing a good performance requires a very accurate torque tracking for the hydraulic robot. With good force sensing in the hydraulic cylinder the torque control can be achieved in constraint motion but not in free motion. A controller using a feedforward term to compensate the velocity related effect is presented and tested experimentally. It is shown that this controller achieves very good results both in free motion and in a constrained task.

Time Domain Finite Element Simulations of Damped Multilayered Beams Using a Prony Series Representation

The dynamic response of steel beams treated with variousconfigurations of constrained viscoelastic layers is consideredexperimentally and analytically. The cantilevered sandwich beams aresubjected to spike inputs and large finite rotations. Frequency domainviscoelastic material properties provided by the manufacturer areconverted to the Prony series time domain expression of the shearrelaxation modulus. A finite element model of the cantilevered beams isimplemented in ABAQUS and numerical predictions are obtained. Naturalfrequencies and damping factors obtained from linear and nonlinear, timedomain finite element analyses are found to be in good agreement withexperimental results.

A Novel Approach for the Dynamic Analysis and Simulation of Constrained Mechanical Systems

In this paper we will outline a formulation for the dynamics of constrained systems. This formulation relies on the D’Alembert-Lagrange principle and the physically meaningful decomposition of the virtual displacements and the generalized velocities for the system where a redundant, non-minimum set of variables is used. The approach is valid for general constrained system with holonomic and/or nonholonomic constraints. We will also discuss a potential application of this formulation in improving the accuracy and stability of the simulation of constrained systems. This application will be demonstrated by an example drawn from space robotics.Copyright

Adaptive Control of Harmonic Drives Based on Virtual Decomposition

Harmonic drives are interesting for robotic applications due to their attractive properties such as high reduction ratio, compact size, low mass, and coaxial assembly. However, the high friction and the dynamics of the flexspline are the main issues that significantly challenge the control systems. In this paper, an adaptive controller capable of adaptively compensating the friction, while incorporating the dynamics of the flexspline, is developed in both joint torque and joint-position control modes. The virtual decomposition control approach allows the dynamics of harmonic drives to be controlled separately from the conventional dynamics of the robots. Adaptive friction compensation and flexspline dynamics based control are the two main contributions of this paper. The L2/Linfin stability and the L 2-gain-induced Hinfin stability are guaranteed. Experimental results demonstrated in both time and frequency domains confirm the feasibility of the proposed approach

Six methods to model a flexible beam rotating in the vertical plane

This paper presents six methods to model a flexible beam rotating in the vertical plane. The first three methods, Hamiltons principle, Lagranges equations for quasi-coordinates and Newton-Euler equations, give a continuous model. The last three methods, Lagranges equations, Jourdains principle and Newton-Euler equations, give a discrete model. The paper discusses how the different methods apply. It also shows how to derive the discrete model directly from the continuous one. The main conclusion of the paper is that if the foreshortening of the flexible link is neglected, then starting from the same assumptions, different methods give different models.

Hardware-in-loop simulation of robots interacting with environment via algebraic differential equation

Hardware-in-the-loop simulation is an attractive tool to validate the functionality of space robots. It is the purpose of the hardware-in-the-loop simulation to make a rigid robot prototype behave, in contact/noncontact phases, as closely as possible to a given robot reference model that would be encountered in the real-world. We introduce a method for the hardware-in-the-loop simulation that is based on the projection of the generalized coordinates of the robot model into two subspaces representing independent and dependent coordinates. The independent coordinate is obtained via simulation by making use of its acceleration model expressed in a closed form. While, the dependent coordinate is obtained from the generalized coordinated of the linearized robot prototype acting as a double integrator on the input dependent-coordinate acceleration. It is followed by investigating the effect of external disturbance on the performance of the hardware-in-the-loop simulation, by the method of Lagrangian multiplier. Finally experimental results obtained from implementation of a single axis arm is presented.

Efficient simulation of a multilayer viscoelastic beam using an equivalent homogeneous beam

Addition of viscoelastic layers on flexible beam improves the behavior of flexible robots. However, the time domain simulation of such multilayer flexible beams is usually done through finite elements which makes it very time consuming. This paper proposes to build an equivalent homogenous flexible beam such that the flexible manipulator can be simulated using the variety of efficient methods developed in robotics. We propose a method to develop the equivalent Voigt-Kelvin model based on the frequency response obtained from the finite-element model. Then, we evaluate the performance of the model by simulating an experimental slewing beam covered with constrained viscoelastic material. The results are very satisfactory and the simulation is one hundred times faster than with the finite element method.