The Grasp Reset Mechanism: An Automated Apparatus for Conducting Grasping Trials

TL;DR

The paper addresses the challenge of collecting large-scale real-world grasp data by introducing the Grasp Reset Mechanism (GRM), an automated apparatus that resets the grasping environment, stabilizes object pose with fixed orientation, and swaps objects without human input. The system combines a mechanical lower reset for positioning and a 3-DOF upper reset for object swapping, alongside a dual-electrical design and a ROS-based software stack with FlexBE for sequence control. It delivers a dataset of 1,020 grasps across four objects using a Kinova Gen3 and Robotiq 2F-85, demonstrating high repeatability (planar position ~0.05 mm, orientation ~2.0 degrees) and substantial real-world data collection efficiency (~17 hours). The GRM and its open-source software enable scalable benchmarking, algorithm validation, and reduced reliance on simulation, with potential as a standard platform for comparing grasping approaches across hardware.

Abstract

Advancing robotic grasping and manipulation requires the ability to test algorithms and/or train learning models on large numbers of grasps. Towards the goal of more advanced grasping, we present the Grasp Reset Mechanism (GRM), a fully automated apparatus for conducting large-scale grasping trials. The GRM automates the process of resetting a grasping environment, repeatably placing an object in a fixed location and controllable 1-D orientation. It also collects data and swaps between multiple objects enabling robust dataset collection with no human intervention. We also present a standardized state machine interface for control, which allows for integration of most manipulators with minimal effort. In addition to the physical design and corresponding software, we include a dataset of 1,020 grasps. The grasps were created with a Kinova Gen3 robot arm and Robotiq 2F-85 Adaptive Gripper to enable training of learning models and to demonstrate the capabilities of the GRM. The dataset includes ranges of grasps conducted across four objects and a variety of orientations. Manipulator states, object pose, video, and grasp success data are provided for every trial.

Figures (7)

• Figure 1: The Grasp Reset Mechanism with a Kinova Gen 3 robot attached. Visible components of interest are labeled.
• Figure 2: A cut-away view of the centering cone and rotation mechanism of the lower reset. Critical components are labeled. A rectangular prism is pictured as an example object.
• Figure 3: A high level view of control of the Grasp Reset Mechanism. The system includes three aspects, the control computer, arm, and GRM itself. The control computer interfaces with the arm via ethernet, WiFi, or USB and the Raspberry Pi on the GRM via ROS over WiFi/Ethernet.
• Figure 4: The FlexBE state machine showing trial progression. A series of trials starts with Test Control in the upper left, then moves down to Trial Control and repeats in a counter-clockwise pattern until all trials are complete (or a failure occurs).
• Figure 5: The magnetic object receptacle is pictured in the bottom left, with three objects from the YCB data set ycb --- a CheezIt box, SPAM can, and spray bottle. Using these objects with the GRM is straightforward, simply a hole must be cut in the bottom and the receptacle glued in.