Group 50
Peter Julian, Oscar Fernandez Paz, Jenny Kim, Sandip Maharaj
Introduction
The goal of this project is to create a robot, "Blue-beary," that can successfully move through an unknown maze without colliding into the walls. The robot should be able to move to certain locations and potentially identify 3-D printed objects which include a teddy bear, book, and cup with the RealSense camera which includes depth sensing. The objects will be identified using YOLO or "You Only Look Once" which is an object detection system. After the objects are successfully identified in the maze, the robot should utilize a 3-D printed actuator with 3 motors to pick up the right object and retrieve it to a certain location.
This project is exciting for several reasons. First, it involves navigating through a maze which uses path planning and obstacle avoidance. The robots need to make adjustments to its velocity depending on how close it gets to the walls of the maze and whether or not it needs to make a turn. Next, it involves object detection which is a step beyond the color detection that was utilized in lab. Also, it includes a 3-D printed actuator that has the ability to scoop up an object from the ground.
The problems that need to be addressed to achieve the end goals include:
Mapping a configurable maze environment
Ensuring that the robot doesn't destroy the maze walls
YOLO and RealSense camera integration
Identifying objects
Creating a working actuator
Real Life Applications
There are many real-world applications that motivated this project. Three potential ideas have been identified below.
Amazon Warehouse (Current Application)
Amazon currently utilizes various robots to redefine how goods are stored and dispatched. In Amazon's warehouses, robots have different tasks that they complete ranging from transporting shelving units to packaging goods. The Kiva robots are similar to our project as it involves finding an object and moving it to the designated location.
2. RescueBot (Future Application)
Currently, when we have dangerous situations like fires or earthquakes, people need to risk their lives to rescue survivors. However, in an ideal world, a self-navigating robot would be able to explore the unknown location or building to find the survivors and carry or escort them to a place of safety. The robot would have the ability to send information to humans and share live video feeds to report the current situation and progress of the mission.
3. Shopping Robot (Every Day Application)
Today, most customers shop for groceries in-person. However, there can be difficulties in the process. For example, the item the shopper is searching for many be too difficult for them to find. Therefore, the use of a shopping robot that assists with in store shopping would be highly useful. This robot would help personal shoppers or other in-store shoppers find the item that they are looking for using AR tags.