Skip to main content

5.3 Manipulation and grasping with humanoid hands

Chapter: Manipulation and Grasping with Humanoid Hands​

🤖 1. Introduction

Humanoid robots ke hands multi‑fingered, high‑DOF (Degrees of Freedom) systems hotay hain jo human-like manipulation perform karte hain. Is chapter mein hum grasping, hand kinematics, control strategies, sensors, aur Isaac Sim integration ko detail mein cover karenge.

Humanoid hand manipulation involves:

  • Object grasping
  • Precision and power grips
  • Finger coordination
  • Sensor‑based feedback
  • Motion planning

🧠 2. Types of Grasping

1. Power Grasp​

Object ko strong force se pakarna.

  • Example: Hammer, water bottle
  • Involved: Palm + all fingers

2. Precision Grasp​

Small object ko fingertips se pakarna.

  • Example: Pen, coin, key

3. Lateral Grasp​

Object ko thumb and side fingers se hold karna.

  • Example: Card holding

4. Tripod Grasp​

Three-finger grip.

  • Example: Small instruments

🦾 3. Humanoid Hand Kinematics

Humanoid hand ke typical configuration:

  • 5 fingers
  • 3–4 joints per finger
  • Tendon‑driven or direct motor-driven actuation
  • ~20 DOF

Finger Motion Structure:​

  • MCP Joint → Flexion/Extension + Abduction
  • PIP Joint → Flexion/Extension
  • DIP Joint → Flexion/Extension

Forward Kinematics:​

Hand pose calculate karta hai based on joint angles.

Inverse Kinematics:​

Finger joint angles compute karta hai taake fingertips target position pe pohanch saken.

🎯 4. Manipulation Workflow Diagram

          +-----------------------------+
          |   Object Detection System   |
          |  (Camera + AI Perception)   |
          +-------------+---------------+
                        |
                        v
          +-----------------------------+
          |  Hand Pose Estimation       |
          | (Target Position & Orientation)|
          +-------------+---------------+
                        |
                        v
          +-----------------------------+
          |  Motion Planning (IK Solver) |
          |  - Finger Angles             |
          |  - Collision Avoidance       |
          +-------------+---------------+
                        |
                        v
          +-----------------------------+
          |  Grasp Execution             |
          |  - Power/Precision Grip      |
          |  - Contact Points Control    |
          +-------------+---------------+
                        |
                        v
          +-----------------------------+
          |   Sensor Feedback Loop       |
          | - Tactile Sensors            |
          | - Force Control              |
          +-----------------------------+

🔍 5. Sensors for Humanoid Manipulation

1. Tactile Sensors​

  • Pressure sensing
  • Slip detection

2. Force/Torque Sensors​

  • Wrist load measurement
  • Over‑gripping prevention

3. Vision Cameras​

  • Object detection
  • Pose estimation

4. IMUs​

  • Hand stability

🧮 6. Grasp Planning Algorithms

✓ Grasp Quality Metrics (GQM)​

Evaluate kartay hain ke grasp stable hai ya nahi.

✓ Common Algorithms:​

  • Antipodal Grasp Detection
  • Grasp Pose CNN (GPCNN)
  • Reinforcement Learning‑based grasping
  • Isaac Sim Hand Pose Generator

🎮 7. Manipulation in Isaac Sim

Isaac Sim provides:

  • Multi‑fingered hand models
  • Physics‑based contact simulation
  • VR and teleoperation support
  • Grasp pose generation

Example (Pseudo Code)​

from omni.isaac.core import World
from omni.isaac.manipulators import Hand

world = World()
hand = world.scene.add(Hand("humanoid_hand"))

# Set target grasp pose
hand.set_target_pose(position, orientation)

# Simulate grasp
while world.is_playing():
    hand.apply_actions()
    world.step()

🏭 8. Real‑World Applications

  • Medical surgery assistants
  • Humanoid service robots
  • Industrial manipulation
  • Space robotics
  • Elderly care assistants

📝 9. Self Assignment

  1. Humanoid hand ka URDF study karein.
  2. Isaac Sim mein ek multi‑finger grasping task setup karein.
  3. Power grasp aur precision grasp ko implement karein.
  4. Tactile sensors ko enable karein aur slip detection test karein.
  5. Report banayein on "Which grasp is more stable and why?"

❓ 10. MCQs

Q1. Precision grasp kis tarah ka grip hota hai?​

A. Whole hand force grip
B. Fingertip grip
C. Wrist rotation
D. Palm-only grip

Q2. Humanoid hand mein MCP joint kis cheez ko control karta hai?​

A. Only rotation
B. Only sideways motion
C. Flexion/Extension + Abduction
D. No movement

Q3. Slip detection kaun sa sensor karta hai?​

A. Camera
B. Tactile sensor
C. Speaker
D. GPS

Q4. Grasp planning algorithm ka main objective kya hota hai?​

A. Robot speed increase karna
B. Stable and secure grip ensure karna
C. Background color change karna
D. Battery life improve karna

Q5. Isaac Sim kis cheez ke liye use hota hai?​

A. 2D cartoon making
B. Physics-based robot simulation
C. Mobile app UI creation
D. Only text editing

✅ Correct Answers

  1. B
  2. C
  3. B
  4. B
  5. B