Congratulations to BIGIDEAS Winners: Isabella Domi, Jessica Palmer, Emery Silberman

April 18, 2017

Congratulations to BIGIDEAS Winners: Isabella Domi, Jessica Palmer, Emery Silberman

Three Mechanical Engineering undergraduates Isabella Domi, Jessica Palmer, Emery Silberman won Berkeley BIGIDEAS competition with $7000 award. Quoting from the award email (4/17/17) "Congratulations! Your Big Ideas project was selected by the judges for a 2nd Place award of $7,000 in the Hardware for Good category. Yours was among a select group of projects awarded from over 320 submissions. We believe your idea has high potential, and we look forward to hearing more about your project as it is implemented!"

This is the 4th award from UC Merced in BIGIDEAS contest history. Past winners from UC Merced are listed here http://bigideas.berkeley.edu/contests/uc-merced/ We are proud to note that, all winners from UC Merced are so far nurtured or incubated in MESA Lab.


Search and Rescue Drone: Using a Trained Neural Network to Detect Missing People
Isabella Domi, Jessica Palmer, Emery Silberman

University of California, Merced

Problem Statement
Current search and rescue methods used in national parks are outdated and inefficient, largely due to a
lack of funding in this department for national parks. In Yosemite National Park, for example, the main
method used in searches involves having YOSAR (Yosemite Search and Rescue) workers and volunteer
responders walk over large areas of land looking for the missing persons and recording on paper the
probability of the persons being in the area surveyed as well as the conditions. If this fails, the second
method includes using the fire department’s helicopter (Helicopter 551) to search for people from an
aerial view, where the team on board the helicopter are trained to identify people from the helicopter [1].
In a search in the Grand Teton National Park for two missing persons in 2011, the SAR team had to lease
a helicopter costing $33,000, with the total cost of the search totalling to an appalling at $115,000 [2].
However after just 51 hours of the person being missing, the chances of survival for that person decreases dramatically [3]. Although using a helicopter to search is more effective over a ground team, visually searching from a helicopter is an incredibly difficult task even for a trained eye. Clearly, these current methods are inefficient and expensive, and therefore not acceptable. With today’s technology, we are capable of both greatly decreasing the search time while increasing the probability of finding missing
persons as well as creating a low-cost solution that can fit into any park’s protocol and budget.
Our idea is to use an unmanned aerial system (UAS) to perform an aerial search over large areas.
Practically anyone can learn how to operate an autonomous UAV and watch the live stream video to
search for people; however to decrease human error our system also uses a trained neural network that can identify people real time from the video feed which provides fast and accurate location data and images on the missing persons.

Team Members

Isabella Domi
Mechanical Engineering and Physics, Undergraduate Student
Licensed sUAS pilot who is capable of collecting data and test flying UAV’s. Also has
contributed to the data processing programming. Has experience with group engineering projects
from working on the SpaceX Hyperloop pod competition, and experience working with UAV’s at
UC Merced’s MESA Lab.

Emery Silberman
Mechanical Engineering, Undergraduate Student
Has prior experience working with neural networks using NVIDIA Digits, and has contributed to
programming the system to process data. Also has years of experience working with model
planes, drones, and autopilot systems.

Jessica Palmer
Mechanical Engineering, Undergraduate Student
Licensed sUAS pilot who has contributed to data collection and ensuring that the UAV’s are
fixed and in good condition to fly. Has experience working with UAV’s from working at MESA,
UC Merced’s own drone research lab.

 

Congratulations!!


Last updated by YangQuan Chen, 4/18/2017.