Sensing and modeling collisions for safer human-robot interaction
Abstract/Contents
- Abstract
- New robotic platforms are quickly being developed. Unlike their predecessors, which were cordoned off from humans, these new robots are designed to work in human centered environments. The goal of these robot platforms is to have robotic technology expand beyond the traditional areas, previously described elsewhere as ``Dirty, Dangerous and Dull, " to new and exciting applications. These new platforms could provide remote users with a physical presence via telepresence technology, provide superior rehabilitative regimens for the injured, or provide the elderly with a high quality independent lifestyle. If the manufacturing sector can provide us with any insight on accident rates, the public should have cause for concern. Accidents are not uncommon, and the injuries can be debilitating. Previous methodologies to develop ``human safe" robots have concentrated on safe behaviors and low-power mechanisms, typically at the expense of reduced performance. However, experts in robot control realize that even the most robust control system cannot cover all possible scenarios. In the case of human robot interaction, control failures could lead to injuries. More generally, performance reductions to insure safety have resulted in platforms that are incapable of completing many desirable tasks. The goal of this dissertation is to develop human safe robot platforms that are capable of performing useful work, while at the same time interacting with humans as safely as possible. Towards this end, the dissertation begins with demonstrating the inadequacy of popular industrial injury metrics: the Chamois Laceration Scale (CLS) and the Head Injury Criterion (HIC). Testing performed at General Motors under the auspices of the Research and Development Division demonstrate the insensitivity of the Chamois Laceration Scale to variations in collision parameters. Similarly, despite the violence of collisions seen during experimentation, the Head Injury Criterion levels achieved in testing were far below those that could produce concussions; robot human collisions do not induce this type of injury because of the energy involved. In response to the lack of appropriate injury measures for robot human collisions, a model based on energy and contact mechanics was developed to map the influence of pre-collision parameters (robot velocity, mass, geometry, approach angle, interface stiffness and friction) on the maximum possible induced tensile stress in a target. Findings demonstrate that normal collisions touted as the ``worst case scenario'' often are not. For an energetically equivalent impactor, an oblique collision can induce significantly higher peak stresses. As such, the importance of interface friction has been underrated. A modeling tool to estimate the worst case stresses in angled collisions has been proposed. To prevent crushing injuries and to sense contact, members of the Biomimetics and Dexterous Manipulation Laboratory developed a capacitive contact sensor. Experiments described in this dissertation demonstrate the efficacy of the sensor in detecting and quantifying collisions. The sensor is effective in both spatial and temporal contact reconstruction. The dissertation ends with suggestions on design and operational guidelines in reducing the likelihood of injury in robot/human collisions.
Description
| Type of resource | text |
|---|---|
| Form | electronic; electronic resource; remote |
| Extent | 1 online resource. |
| Publication date | 2013 |
| Issuance | monographic |
| Language | English |
Creators/Contributors
| Associated with | Phan, Samson |
|---|---|
| Associated with | Stanford University, Department of Mechanical Engineering. |
| Primary advisor | Cutkosky, Mark R |
| Thesis advisor | Cutkosky, Mark R |
| Thesis advisor | Delp, Scott |
| Thesis advisor | Levenston, Marc Elliot |
| Advisor | Delp, Scott |
| Advisor | Levenston, Marc Elliot |
Subjects
| Genre | Theses |
|---|
Bibliographic information
| Statement of responsibility | Samson Phan. |
|---|---|
| Note | Submitted to the Department of Mechanical Engineering. |
| Thesis | Thesis (Ph.D.)--Stanford University, 2013. |
| Location | electronic resource |
Access conditions
- Copyright
- © 2013 by Samson Phan
- License
- This work is licensed under a Creative Commons Attribution Non Commercial No Derivatives 3.0 Unported license (CC BY-NC-ND).
Also listed in
Loading usage metrics...