Jump to content

Connect SuperML | Leeroopedia MCP: Equip your AI agents with best practices, code verification, and debugging knowledge. Powered by Leeroo — building Organizational Superintelligence. Contact us at founders@leeroo.com.

Implementation:ARISE Initiative Robosuite RobotiqThreeFingerGripper

From Leeroopedia
Knowledge Sources
Domains Robotics, Simulation, MJCF_Modeling
Last Updated 2026-02-15 07:00 GMT

Overview

RobotiqThreeFingerGripper provides the Robotiq 3-Finger Adaptive Gripper model with 11-DOF base actuation, plus 1-DOF synchronized and 4-DOF dexterous variants.

Description

This module defines three classes forming a hierarchy of Robotiq three-finger gripper models. RobotiqThreeFingerGripperBase loads the MJCF XML from grippers/robotiq_gripper_s.xml and defines 11 initial joint positions (all zeros). The collision geometry groups two fingers (f1 and f2) under left_finger with segments l0-l3, tip, and pad for each, and the third finger (f3) under right_finger with l0-l3, tip, and pad geometries.

RobotiqThreeFingerGripper provides 1-DOF synchronized control where a single scalar action drives all three fingers together. It uses the formula: current_action = clip(current_action + speed * action, -1, 1), then outputs a 4-element array by concatenating current_action * [1, 1, 1] with [-1]. The appended -1 automatically sets the scissor joint (which controls the spread between fingers 1 and 2) to its closed position. The speed is 0.2.

RobotiqThreeFingerDexterousGripper provides 4-DOF independent control, where each of the three fingers and the scissor joint can be actuated independently. It uses the formula: current_action = clip(current_action + speed * sign(action), -1, 1). This variant enables advanced grasping strategies that exploit the adaptive capabilities of the three-finger design, such as enveloping grasps and precision pinches.

Usage

Use RobotiqThreeFingerGripper (1-DOF) for standard manipulation tasks requiring robust three-finger grasping. Use RobotiqThreeFingerDexterousGripper (4-DOF) for advanced manipulation requiring independent finger and scissor joint control. The base class provides full 11-DOF raw actuator access for maximum flexibility.

Code Reference

Source Location

Signature

class RobotiqThreeFingerGripperBase(GripperModel):
    def __init__(self, idn=0):

class RobotiqThreeFingerGripper(RobotiqThreeFingerGripperBase):
    # Inherits __init__; overrides format_action, speed, dof

class RobotiqThreeFingerDexterousGripper(RobotiqThreeFingerGripperBase):
    # Inherits __init__; overrides format_action, speed, dof

Import

from robosuite.models.grippers.robotiq_three_finger_gripper import (
    RobotiqThreeFingerGripper,
    RobotiqThreeFingerGripperBase,
    RobotiqThreeFingerDexterousGripper,
)

I/O Contract

Inputs

Name Type Required Description
idn int or str No Unique identification number or string for this gripper instance. Default: 0

Outputs

Name Type Description
RobotiqThreeFingerGripperBase instance GripperModel Base 11-DOF three-finger gripper model
RobotiqThreeFingerGripper instance GripperModel 1-DOF synchronized three-finger gripper with auto-closed scissor (speed: 0.2)
RobotiqThreeFingerDexterousGripper instance GripperModel 4-DOF independent finger and scissor gripper (speed: 0.2)
init_qpos np.array Array of 11 zeros for default joint positions

Usage Examples

from robosuite.models.grippers.robotiq_three_finger_gripper import (
    RobotiqThreeFingerGripper,
    RobotiqThreeFingerDexterousGripper,
)
import numpy as np

# 1-DOF synchronized gripper (scissor auto-closed)
gripper = RobotiqThreeFingerGripper(idn=0)
action = np.array([1.0])  # close all fingers
formatted = gripper.format_action(action)  # 4-element: [close, close, close, -1]

# 4-DOF dexterous gripper (3 fingers + scissor)
dex_gripper = RobotiqThreeFingerDexterousGripper(idn=0)
action = np.array([1.0, -1.0, 0.5, 0.0])  # independent control
formatted = dex_gripper.format_action(action)  # 4-element independent output

print(gripper.init_qpos)  # array of 11 zeros
print(gripper.dof)  # 1
print(dex_gripper.dof)  # 4

Related Pages

Page Connections

Double-click a node to navigate. Hold to expand connections.
Principle
Implementation
Heuristic
Environment