Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

The Door and Drawer Reset Mechanisms: Automated Mechanisms for Testing and Data Collection

Kyle DuFrene, Luke Strohbehn, Keegan Nave, Ravi Balasubramanian, Cindy Grimm

TL;DR

Door and drawer manipulation in real environments is data-hungry and highly variable, hindering benchmarking of control policies. The authors present two automated, open-source testbeds, DORM and DWRM, that enable repeatable testing with variable resistance and sensor feedback, built at under $400 each. They pair the hardware with software for automated data collection and a dataset of 660 trials (over 600 grasps) across two Kinova manipulators and two pull attachments, capturing multimodal data including video, sensor, and wrench-like forces. The work provides a practical platform for benchmarking hardware and control algorithms and highlights real-world noise that should be accounted for in simulation and learning.

Abstract

Robotic manipulation in human environments is a challenging problem for researchers and industry alike. In particular, opening doors/drawers can be challenging for robots, as the size, shape, actuation and required force is variable. Because of this, it can be difficult to collect large real-world datasets and to benchmark different control algorithms on the same hardware. In this paper we present two automated testbeds, the Door Reset Mechanism (DORM) and Drawer Reset Mechanism (DWRM), for the purpose of real world testing and data collection. These devices are low-cost, are sensorized, operate with customized variable resistance, and come with open source software. Additionally, we provide a dataset of over 600 grasps using the DORM and DWRM. We use this dataset to highlight how much variability can exist even with the same trial on the same hardware. This data can also serve as a source for real-world noise in simulation environments.

The Door and Drawer Reset Mechanisms: Automated Mechanisms for Testing and Data Collection

TL;DR

Door and drawer manipulation in real environments is data-hungry and highly variable, hindering benchmarking of control policies. The authors present two automated, open-source testbeds, DORM and DWRM, that enable repeatable testing with variable resistance and sensor feedback, built at under $400 each. They pair the hardware with software for automated data collection and a dataset of 660 trials (over 600 grasps) across two Kinova manipulators and two pull attachments, capturing multimodal data including video, sensor, and wrench-like forces. The work provides a practical platform for benchmarking hardware and control algorithms and highlights real-world noise that should be accounted for in simulation and learning.

Abstract

Robotic manipulation in human environments is a challenging problem for researchers and industry alike. In particular, opening doors/drawers can be challenging for robots, as the size, shape, actuation and required force is variable. Because of this, it can be difficult to collect large real-world datasets and to benchmark different control algorithms on the same hardware. In this paper we present two automated testbeds, the Door Reset Mechanism (DORM) and Drawer Reset Mechanism (DWRM), for the purpose of real world testing and data collection. These devices are low-cost, are sensorized, operate with customized variable resistance, and come with open source software. Additionally, we provide a dataset of over 600 grasps using the DORM and DWRM. We use this dataset to highlight how much variability can exist even with the same trial on the same hardware. This data can also serve as a source for real-world noise in simulation environments.
Paper Structure (13 sections, 6 figures, 2 tables)

This paper contains 13 sections, 6 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. The Mechanisms
  4. Mechanical Design
  5. Door
  6. Drawer
  7. Pull Attachments
  8. Electrical Design
  9. Software
  10. Dataset
  11. Dataset Methods
  12. Dataset Results
  13. Discussion and Conclusions

Figures (6)

  • Figure 1: The Drawer Reset Mechanism with a Kinova JACO 7-dof arm. As pictured, the instrumented knob is attached to the Drawer Reset Mechanism.
  • Figure 2: Pictured on the left is the Door Reset Mechanism, on the right is the Drawer Reset Mechanism. Important parts are labeled, including electronics, sensors, motors, and pull attachments.
  • Figure 3: The knob (left) and handle (right) used as the pull attachments for the dataset in this paper. Each have a silicone wrapper to protect and distribute grasping forces to the force sensitive resistors (FSRs). The silicone on the handle is rolled back to show the FSRs.
  • Figure 4: The FlexBE state machine showing trial progression for standard grasping trials with the DORM and DWRM.
  • Figure 5: Our visualisation interface showing a trial. Buttons, drop-downs, and text boxes allow user inputs to enable or disable certain features. Four visualization windows are displayed. Top left: a view of the Kinova JACO 2 states, Top right: a 3D model of the handle, Bottom left: a graph of the drawer distance recorded by the ToF, Bottom right: a plot of FSR data.
  • ...and 1 more figures