Establishing a Comprehensive Database for Safe Human-Robot Interaction with Edged or Pointed Objects

Core Concepts

This study establishes an initial database of human hand injury patterns caused by impacts with edged or pointed objects, using pig dew claws and chicken drumsticks as surrogates. The generated datasets enable efficient risk assessment and safe integration of such tools or objects in physical human-robot interaction applications.

Abstract

This study aims to create a comprehensive human hand injury protection database to enable efficient risk assessment and safe integration of pointed or edged tools or objects in physical human-robot interaction (pHRI).
The researchers used pig dew claws and chicken drumsticks as surrogates for the human hand to conduct 351 and 117 impact experiments, respectively, with three different impactor geometries (wedge, edge, and sheet). The observed injury patterns were classified and documented, including skin imprints, cuts, muscle injuries, and bone injuries.
For the pig dew claws, skin imprints and cuts were the predominant observations across all impactors. Bone injuries occurred with the sheet impactor at velocities above 1.0 m/s and masses of 2.6 kg or more. The edge impactor also caused muscle injuries at velocities above 1.0 m/s and masses of 0.6 kg or more.
In contrast, the chicken drumsticks showed more muscle and bone injuries, with skin imprints and cuts only observed in the distal impact location. The researchers linked the measured contact forces to the observed injury types, providing a reference for safe impact scenarios.
To validate the applicability of the generated datasets, the researchers conducted experiments with a robot performing constrained contacts with a Phillips head screwdriver and a breadboard, demonstrating that safe velocities can be determined based on the injury prevention datasets to avoid open skin, muscle, or bone injuries.
The study highlights the importance of considering edged and pointed geometries in pHRI safety, as they cannot be completely avoided in real-world applications. The generated datasets provide a foundation for efficient risk assessment and safe integration of such tools or objects, paving the way for further collaborative efforts to create a comprehensive human injury avoidance database for any pHRI scenario.

Stats

The maximum velocity applied in the validation study was 0.2 m/s.
The effective mass at the point of interest was 1.50 kg for the Phillips head scenario and 1.92 kg for the breadboard scenario.

Quotes

None

Key Insights Distilled From

Towards Safe Robot Use with Edged or Pointed Objects

by Robin Jeanne... at arxiv.org 04-08-2024

https://arxiv.org/pdf/2404.04004.pdf

Deeper Inquiries

How can the generated datasets be extended to account for shearing contacts and unconstrained impact scenarios

To extend the generated datasets to account for shearing contacts and unconstrained impact scenarios, additional experiments can be conducted using different impactor geometries and contact conditions. For shearing contacts, impactors with specific shapes and orientations can be used to simulate scenarios where the force is applied parallel to the skin surface, leading to shearing injuries. By varying the angle and direction of the impact, the dataset can capture the injury patterns resulting from shearing forces.
For unconstrained impact scenarios, where the contact is not limited or constrained, the experiments can be designed to simulate a wider range of real-world interactions. This can involve using different impact velocities, masses, and impactor geometries to mimic scenarios where the hand is exposed to various types of impacts without any restrictions. By systematically varying these parameters and observing the resulting injuries, the dataset can be expanded to include a comprehensive range of injury patterns that may occur in unconstrained impact situations.

What are the limitations of using animal surrogates, and how can the findings be validated with human cadaver or in-vivo studies

Using animal surrogates in injury studies has certain limitations that need to be addressed for the findings to be validated and extrapolated to human scenarios. One limitation is the anatomical differences between animal tissues and human tissues, which may affect the injury patterns observed in the experiments. Additionally, the biomechanical properties of animal tissues may not fully represent those of human tissues, leading to discrepancies in the injury responses.
To validate the findings from animal surrogate studies, further experiments can be conducted using human cadaver specimens or in-vivo studies. Human cadaver studies allow for a more direct comparison of injury responses between animal and human tissues, providing a more accurate representation of potential injury patterns. In-vivo studies on human subjects can also validate the findings and provide insights into the actual injury mechanisms and responses in real-life scenarios.

What other factors, such as age, gender, or pre-existing conditions, could influence the observed injury patterns, and how can they be incorporated into the database

Several factors, such as age, gender, and pre-existing conditions, can influence the observed injury patterns in human-robot interaction scenarios. Age-related changes in tissue properties, such as skin elasticity and bone density, can impact the susceptibility to injuries. Gender differences in tissue composition and structure may also play a role in determining the severity and type of injuries sustained.
Incorporating these factors into the injury database requires conducting experiments on a diverse range of subjects that represent different age groups, genders, and health conditions. By systematically varying these factors in the experiments and analyzing the injury responses, the database can capture the influence of age, gender, and pre-existing conditions on the observed injury patterns. This comprehensive approach will provide a more nuanced understanding of how these factors contribute to the overall risk assessment in human-robot interaction scenarios.