An official website of the United States government
Abstract This paper explores the use of Gaussian process regression for system identification in control engineering. It introduces two novel approaches that utilize the data from a measured global system error. The paper demonstrates these approaches by identifying a simulated system with three subsystems, a one degree of freedom mass with two antagonist muscles. The first approach uses this whole-system error data alone, achieving accuracy on the same order of magnitude as subsystem-specific data ( of total model errors). This is significant, as it shows that the same data set can be used to identify unique subsystems, as opposed to requiring a set of data descriptive of only a single subsystem. The second approach demonstrated in this paper mixes traditional subsystem-specific data with the whole system error data, achieving up to 98.71% model improvement.
more »
« less
Non-expert Caregivers to Improve the Identification of a Physiologically Actuated Robot
Functional electrical stimulation is a promising technique for restoring arm function to those with paralysis from a high spinal cord injury. While simple controllers are easy to implement, model-based controllers are likely better equipped to leverage the arm’s kinematic and dynamic complexity, particularly for the high variations associated with functional arm movement. One modelling technique for a model-based controller is Gaussian Process Regression. Previous simulation work has shown promise leveraging whole-arm error data to identify the arm’s various subsystems, but used perfect simulated data. We asked caregivers to correct a robotic arm’s movement as simulated muscles generated torque. The simulated muscles were controlled as if they were electrically stimulated human arm muscles. This study demonstrates non-expert caregivers’ ability to collect this error data via whole-arm corrections, and provides insight into their ability to improve arm subsystem models made with Gaussian Process Regression. Despite significant error in caregivers’ ability to provide force corrections to hold the robot in a static configuration, these corrections were leveraged to significantly improve muscle models; the muscles that improved the most were the ones primarily used to move the physiologically actuated robot.
more »
« less
- PAR ID:
- 10587415
- Publisher / Repository:
- IEEE
- Date Published:
- ISBN:
- 979-8-3503-7502-2
- Page Range / eLocation ID:
- 1477 to 1482
- Format(s):
- Medium: X
- Location:
- Pasadena, CA, USA
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Abstract In this study, a 13 degrees of freedom (DOFs) three-dimensional (3D) human arm model and a 10 DOFs 3D robotic arm model are used to validate the grasping force for human-robot lifting motion prediction. The human arm and robotic arm are modeled in Denavit-Hartenberg (DH) representation. In addition, the 3D box is modeled as a floating-base rigid body with 6 global DOFs. The human-box and robot-box interactions are characterized as a collection of grasping forces. The joint torque squares of human arm and robot arm are minimized subjected to physics and task constraints. The design variables include (1) control points of cubic B-splines of joint angle profiles of the human arm, robotic arm, and box; and (2) the discretized grasping forces during lifting. Both numerical and experimental human-robot liftings were performed with a 2 kg box. The simulation reports the human arm’s joint angle profiles, joint torque profiles, and grasping force profiles. The comparisons of the joint angle profiles and grasping force profiles between experiment and simulation are presented. The simulated joint angle profiles have similar trends to the experimental data. It is concluded that human and robot share the load during lifting process, and the predicted human grasping force matches the measured experimental grasping force reasonably well.more » « less
-
In an effort to combine two rehabilitation strategies, Functional Electrical Stimulation (FES) and robotic therapy, a rehabilitation robot was developed to challenge an arm during bicep curls elicited by closed-loop control of FES. The robot is designed to act as an admittance and its robust, sliding mode controller is proven to be passive with respect to the human. The FES controller utilizes a robust, sliding mode control design to then dominate the robot effects and obtain global exponential stability as demonstrated by a Lyapunov-based stability analysis. The two interacting controllers yield arm position and velocity regulation, where the robot challenges this movement with a desired admittance.more » « less
-
Ab initio microkinetic modeling, parameterized using density functional theory (DFT) energies, is a common tool to quantify reaction rates and analyze reaction mechanisms a priori in heterogeneous catalysis. Such models, however, often have large prediction errors even if they include plausible reaction steps and correctly model the active sites; this is partially due to the intrinsic inaccuracies of the chosen DFT functional. Borrowing concepts from Bayesian calibration theory, we show that transferable data-driven corrections to DFT energies in the form of Gaussian process models trained on single-crystal adsorption calorimetry data can improve the accuracy of microkinetic models substantially. Specifically, we demonstrate that such corrections improve the predictive accuracy of the microkinetic model of the water-gas shift reaction on single-crystal Cu(111) surface by 3 orders of magnitude. We finally show that Gaussian process corrections serve as informed priors in a Bayesian experimental design framework to learn an accurate a posteriori microkinetic model from few kinetic experiments. We posit that these results suggest that even infusing small, related, high-fidelity thermochemistry data, when available, can systematically and substantially improve the predictive accuracy of microkinetic models.more » « less
-
Effective physical human-robot interaction (pHRI) depends on how humans can communicate their intentions for movement with others. While it is speculated that small interaction forces contain significant information to convey the specific movement intention of physical humanhuman interaction (pHHI), the underlying mechanism for humans to infer intention from such small forces is largely unknown. The hypothesis in this work is that the sensitivity to a small interaction force applied at the hand is affected by the movement of the arm that is affected by the arm stiffness. For this, a haptic robot was used to provide the endpoint interaction forces to the arm of seated human participants. They were asked to determine one of the four directions of the applied robot interaction force without visual feedback. Variations of levels of interaction force as well as arm muscle contraction were applied. The results imply that human’s ability to identify and respond to the correct direction of small interaction forces was lower when the alignment of human arm movement with respect to the force direction was higher. In addition, the sensitivity to the direction of the small interaction force was high when the arm stiffness was low. It is also speculated that humans lower their arm stiffness to be more sensitive to smaller interaction forces. These results will help develop human-like pHRI systems for various applications.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Conference Paper:
- https://doi.org/10.1109/RO-MAN60168.2024.10731279
