Discipline: Physics
Subcategory: Physics (not Nanoscience)
Session: 4
Lewis Campbell - Morehouse College
Co-Author(s): Kehinde Aina, Georgia Institute of Technology, School of Physics Bahnisihka Dutta, Georgia Institute of Technology, School of Physics Daniel I. Goldman , Georgia Institute of Technology, School of Physics.
The performance of robots when responding to disasters is affected by issues experienced with the terrain or with the environment of the disaster site. For example, the Carnegie Mellon University Snake robot experienced difficulties navigating the terrain of the Mexico City earthquake disasters and in locating victims1. This can be attributed to the fact that the field conditions are not regulated as much as they are in lab conditions or wherever the robot is created and tested. Hopefully, by observing a robot’s performance over time it could be possible to determine the optimal conditions for performance, but more importantly, the conditions in which the robot starts breaking down or losing functionality when deployed in disaster situations based off these parameters.
For the purpose of this research project, we studied the ant excavation robots and discover metrics that can be used to evaluate the maximum amount of stress the robots can handle. Stress, in this case, relates to the sensory damage the robot experiences, caused by a repeated collision with external objects. There are more variables that factor into this, but the collisions are the most apparent form of stress the robots undergo. We began the process of differentiating the types of contact to observe the correlation between both frequency and type of contact and sensory performance. To find the threshold to differentiate ant and wall vs. ant to ant contact the robots were placed in a confined area in which they collided with the walls lined with aluminum tape and then logged the serial outputs. Then the ants were removed from the area and allowed to collide with each other making contact between the copper tape on the sides of the robots. The outputs were then graphed with a moving average trendline to determine if there was an acceptable threshold that signified contact with a wall or ant. The graphed data demonstrated that the method used to differentiate contact was not effective, and due to the limited time available the most efficient way to differentiate contact between walls and ants. The research will continue with testing new mediums to differentiate contact type, then experimentation will develop to begin observing the effect of the frequency of contacts.
Funder Acknowledgement(s): NSF Grant# 1560165
Faculty Advisor: Dr. Dan Goldman, LewFitz725@gmail.com
Role: I completed all of the research myself with help from my mentors and coauthors.