Analysis of human grasping behavior: object characteristics and grasp type

Area of Science:

• Human-computer interaction
• Robotics
• Biomechanics

Background:

• Understanding human grasping is crucial for designing intuitive robotic systems and effective rehabilitation tools.
• Previous research has not comprehensively analyzed the physical properties of objects during naturalistic grasping tasks.

Purpose of the Study:

• To analyze human grasping behavior in unstructured tasks.
• To correlate grasp types with object properties.
• To inform the design of robotic manipulators, hand rehabilitation, and haptic devices.

Main Methods:

• Analysis of nearly 10,000 grasp instances from two housekeepers and two machinists.
• Classification of grasped objects based on seven properties: mass, shape, size of grasp location, grasped dimension, rigidity, and roundness.

Main Results:

• 55% of grasped objects exceeded 15 cm in at least one dimension, indicating limitations in grasping by the largest axis.
• 92% of objects weighed 500g or less, suggesting high payload capacity is often not required.
• 96% of grasp locations were 7 cm or less in width, providing data for grasp aperture design.

Conclusions:

• Human grasping behavior favors grasping the smallest object dimension (94% of instances).
• This suggests a default strategy for grasp planners in robotic systems.
• Findings provide critical data for optimizing robotic hand design and hand rehabilitation programs.