Derek Hanely, (junior, mathematics and economics), spent his summer in Fairmont, W. Va., working with a team of interns at the Independent Verification and Validation Facility.
According to Hanely, there, NASA employs “double triple checkers” to analyze mission-critical software that can tolerate minimum error at the risk of the failure of multi-billion dollar projects.
Under the mentorship of Jeremy Yagle, a recent graduate of Indiana University of Pennsylvania’s applied mathematics graduate program, Hanely, an applied statistics minor, and four other interns worked on Robotics Capabilities Development initiative.
The five summer interns specifically researched Computer Vision, the ability of a computerized robot to “see” objects, shapes or landscapes and features thereof via camera and recognize them as a human would.
This technology will enable NASA to service or refuel satellites via crafts with robotic control capabilities, redirect asteroids from hitting Earth, collect samples from asteroids for scientific testing and eventually collect enough data to put humans on Mars as per the Mars 2020 Rover Missions.
Hanely and his team spent 10 weeks at IV&V assessing and analyzing existing algorithms for image processing – the manipulation of image
inputs via specific algorithms.
For example, some algorithms may program the computer to detect the edges of objects and planes using intricate calculus and thresh holding.
A different algorithm pre-processes an image to get it ready for the computer to analyze and use.
Still, another algorithm “smoothes” out images to highlight only the most important parts and erase extraneous or erroneous details.
A person could command a computer fitted with this technology to locate every object’s edge in a picture.
After becoming familiar with C++ open-source software platforms, Hanely was primarily assigned a two-fold job.
First, he had to “read,” analyze and understand approximately 30 such image-processing algorithms.
Then, he had to characterize them in such a way that an analyst could understand the goals, risks, ideal environment and limitations of certain programs.
Once written, these documents were filed into a technical reference database accessible and comprehensible by any analyst.
Other members of Hanely’s team studied the integration of robot-arm control with the image processing software.
The students used inverse kinematics to calculate the appropriate joint angles required for the robotic arm to grasp onto a particular feature of an object identified via computer vision.
Hanely collaborated with these students to help him characterize the algorithms’ risks and problem areas.
Complications occasionally arise in image processing algorithms because of light conditions and radiation, which can distort an image or damage a lens.
Just as the human eye sometimes has difficulty identifying details on an object if it is lit from the back, image processing software experiences similar troubles.
Hanely and his team identified some of these problem areas in the algorithms and suggested timing or corrections to circumvent these issues.
Their findings and analyses of the “operational envelope,” or the conditions under which the algorithms do and don’t work, were also documented with Hanely’s work.
Using this work in computer vision and much deeper research into the body’s reaction to zero-gravity long-term, NASA has set ambitious goals.
By as soon as 2030, NASA hopes to send a manned mission to the planet Mars.
Additionally, NASA anticipates having a completely autonomous service system for its satellites and rovers.
There will be no need for a man with a joystick on Earth because computer vision will enable computerized robots to repair the machines on their own.
With this internship under his belt, Hanely hopes to follow the path of Yagle, enter a cooperative program with NASA and continue studying applied mathematics.