Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Fast Payload Calibration for Sensorless Contact Estimation Using Model Pre-training

Shilin Shan, Quang-Cuong Pham

TL;DR

This work tackles sensorless contact estimation under payload variations by introducing offline pre trained neural network models that cover a wide joint space. The authors propose two architectures, PsPM and PaPM, to enable fast online calibration with only a 4 s trajectory by leveraging extensive offline data collected across 69 discretized payloads. PaPM integrates a Payload Indicator to consolidate payload information into a single model, reducing online compute and memory needs while maintaining high accuracy across configurations; PsPM uses multiple payload specific models for improved accuracy. The approach demonstrates improved joint space and task space compliance and reliable wrench estimation, offering practical benefits for rapid recalibration in dynamic manipulation tasks with varying payloads. Overall, the method enables robust, fast dynamics adaptation crucial for industrial manipulation and collaborative robotics.

Abstract

Force and torque sensing is crucial in robotic manipulation across both collaborative and industrial settings. Traditional methods for dynamics identification enable the detection and control of external forces and torques without the need for costly sensors. However, these approaches show limitations in scenarios where robot dynamics, particularly the end-effector payload, are subject to changes. Moreover, existing calibration techniques face trade-offs between efficiency and accuracy due to concerns over joint space coverage. In this paper, we introduce a calibration scheme that leverages pre-trained Neural Network models to learn calibrated dynamics across a wide range of joint space in advance. This offline learning strategy significantly reduces the need for online data collection, whether for selection of the optimal model or identification of payload features, necessitating merely a 4-second trajectory for online calibration. This method is particularly effective in tasks that require frequent dynamics recalibration for precise contact estimation. We further demonstrate the efficacy of this approach through applications in sensorless joint and task compliance, accounting for payload variability.

Fast Payload Calibration for Sensorless Contact Estimation Using Model Pre-training

TL;DR

This work tackles sensorless contact estimation under payload variations by introducing offline pre trained neural network models that cover a wide joint space. The authors propose two architectures, PsPM and PaPM, to enable fast online calibration with only a 4 s trajectory by leveraging extensive offline data collected across 69 discretized payloads. PaPM integrates a Payload Indicator to consolidate payload information into a single model, reducing online compute and memory needs while maintaining high accuracy across configurations; PsPM uses multiple payload specific models for improved accuracy. The approach demonstrates improved joint space and task space compliance and reliable wrench estimation, offering practical benefits for rapid recalibration in dynamic manipulation tasks with varying payloads. Overall, the method enables robust, fast dynamics adaptation crucial for industrial manipulation and collaborative robotics.

Abstract

Force and torque sensing is crucial in robotic manipulation across both collaborative and industrial settings. Traditional methods for dynamics identification enable the detection and control of external forces and torques without the need for costly sensors. However, these approaches show limitations in scenarios where robot dynamics, particularly the end-effector payload, are subject to changes. Moreover, existing calibration techniques face trade-offs between efficiency and accuracy due to concerns over joint space coverage. In this paper, we introduce a calibration scheme that leverages pre-trained Neural Network models to learn calibrated dynamics across a wide range of joint space in advance. This offline learning strategy significantly reduces the need for online data collection, whether for selection of the optimal model or identification of payload features, necessitating merely a 4-second trajectory for online calibration. This method is particularly effective in tasks that require frequent dynamics recalibration for precise contact estimation. We further demonstrate the efficacy of this approach through applications in sensorless joint and task compliance, accounting for payload variability.
Paper Structure (18 sections, 10 equations, 10 figures, 4 tables)

This paper contains 18 sections, 10 equations, 10 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Calibration Methods
  4. Model-based Method
  5. Online Learning Model (OLM)
  6. Payload-specific Pre-trained Models (PsPM)
  7. Payload-adaptive Pre-trained Model (PaPM)
  8. Training Data Collection
  9. Data Collection for Offline Learning
  10. Payload Discretization
  11. Trajectory Excitation
  12. Data Size Investigation
  13. Implementation of PI and Calibration Trajectory
  14. Model Training and Comparisons
  15. Application
  16. ...and 3 more sections

Figures (10)

  • Figure 1: Snapshot of the robot compliance task. A video demonstration is available as the supplementary or at: https://youtu.be/r5SYtfSI6uc
  • Figure 2: The block diagram and data flow for OLM. The implication of the variables can be found in Table \ref{['Table:Var_Imp']}.
  • Figure 3: The block diagram and data flow for PsPM. The implication of the variables can be found in Table \ref{['Table:Var_Imp']}.
  • Figure 4: The block diagram and data flow for PaPM. The implication of the variables can be found in Table \ref{['Table:Var_Imp']}.
  • Figure 5: The end-effector design for payload adjustment.
  • ...and 5 more figures