Analyzing unidentified locked-joint failures in kinematically redundant manipulators

Abstract

Robots are frequently used for operations in hostile environments. The very nature of these environments, however, increases the likelihood of robot failures. Common failuretolerance techniques rely on effective failure detection and identification. Since a failure may not always be successfully identified, or, even if identified, may not be identified soon enough, it becomes important to consider the behavior of manipulators with unidentified failures. This work investigates the behavior of robots experiencing unidentified locked-joint failures in a general class of tasks characterized by point-to-point motion. Based on the analysis, a procedure for workspace evaluation is developed that allows for the identification of regions in the manipulator's workspace in which tasks may be completed even with such failures. where xe R" is the position of the end effector, q eRn is the vector of joint variables, and m and n the dimensions of the task space and joint space, teepeetively. Manipulators that have more degrees of freedom (DOFs) than required for a task, i.e., n>m, are said to be redundant. 'The end-effector velocity is expressed in terms of the joint rates as where] E H m X n is the manipulator Jacobian, xis the end-effector velocity, and it is the joint velocity.If perfect servo control of the joints is assumed, then in a healthy manipulator the actual joint velodfailed, the ability of manipulators to converge to desired end-effector positions even with imperfect/ approximated lacobtana/Iacobian-inverses has been addressed in refs. 26 and 27. Therefore, we focus on the other case where a joint has actually locked, but the controller continues to command motion of that joint as though it were healthy. For a general class of tasks characterized by point-to-point motion, we examine convergence issues such as whether the manipulator comes to rest and, if so, what is the terminal position and orientation of the end-effector. The convergence analysis is restricted to purely kinematic effects, due to the fact that the dynamic effects of a failure are essentially transient in nature and do not significantly affect the convergence behavior. 18.28 -3 0 Conditions under which the manipulator converges are explicitly defined and the anomalies in behavior due to the faults explained and illustrated with examples. These conditions are used to evaluate the convergence behavior of the manipulator over the postfailure workspace. This allows for the identification of workspace regions for which task completion may be possible even with unidentified failures.The position and orientation-of the end effector of a manipulator can be expressed in terms of its joint variables by the kinematic equation "nus assumes a typical commercial robot where duplicate sensing and/or analytical redundancy does not exist.

A fault detection and isolation system for cooperative manipulators

“…The FDI system is proposed based on the following faults: FSJF, where an actuation loss occurs in one arm joint (English and Maciejewski, 1998); LJF, where one arm joint is locked (Goel et al, 2004); JPF, where the measurement of the joint position is not correct (Notash, 2000), and JVF, where the measurement of the joint velocity is not correct. JPFs and JVFs generally occur due to sensor faults.…”

Thermodynamic analysis of a space station remote manipulator with a harmonic drive that considers an integrated thermal protection layer

“…Therefore providing meaningful predictions of thermodynamic behavior is critical to determine thermal protection strategies for SSRMs during the early design stage [6,7]. Thermally induced deformations due to sustained dynamic heating can jointly affect the performance characteristics of space manipulator, such as ''locked joint,'' where the affected joint's velocity is identically zero [8]. This will lead to the energy consumption, generated torque and speed fluctuations.…”

QR Code-Based Self-Calibration for a Fault-Tolerant Industrial Robot Arm

“…To the best of our knowledge, there are mainly two approaches for robots to adapt to malfunctions: model-based and learning-based. Works such as the one presented by [6] tackle the problem from a mathematical perspective, while others, such as [7], [8] and [9] use heuristics to converge to a solution.…”

“…From a model-based perspective, Goel et al [6] focus on a case where one of the joints is locked but the controller continues to control that joint as though it were healthy. For a general class of tasks characterized by pointto-point motion, they examine convergence issues such as whether the manipulator comes to rest and, if so, what is the terminal position and orientation of the end-effector.…”

“…A great body of research has been combining learning algorithms with robotics, and some focus on applications on robot manipulators: Jutharee and Maneewarn [7] use a Genetic Algorithm optimization method to the 7 degree-offreedom (DOF) of a semi-humanoid robot when a joint failure occurred. However, similarly to [6], their work assumes that the constrained angle value is known. Lou et al [12] propose a PS-PCA algorithm to improve the fault detection rate.…”