Solution Procedure
Collection:
In order to pick up two types of trash (represented as a soda can and golf balls) FRUSTRATION incorporates two capturing mechanisms. The primary trash collection mechanism is a series of passive gates that lock the trash into one of two storage bins as FRUSTRATION drives over the trash. The two bins correspond to recyclable (blue) and non-recyclable (orange). Additionally, FRUSTRATION incorporates a powerful gripper which will pick up and deposit any trash too great to be swept underneath. These two trash collection methods give FRUSTRATION unparalleled versatility.
Detection:
FRUSTRATION incorporates a number of sensors which enable it to navigate the demonstration course and identify trash successfully. Three Infrared Light sensors are used to determine the distance from the robot and walls in front of and to the left or right of the robot. Encoders on the wheels are used to measure wheel speed and a gyro is used to measure the robot’s heading. The information provided by these sensors is used to determine to robot’s position. FRUSTRATION also incorporates a camera whose image is processed using segmentation code to identify different types of trash and distinguish recyclable (blue) from non-recyclable (orange) trash. Finally, a compass is used to give the robot’s orientation relative to an absolute coordinate system.
Logic:
The coding used to
give FRUSTRATION intelligence and awareness is a state machine with 12
states. It begins by searching for
trash too large to be swept underneath (soda can) and once it is identified,
the robot will head towards the object.
This is accomplished with the camera.
Once the image of the object falls off of the camera, the gripper is
closed and the large trash is stored within FRUSTRATION. Then the 2nd state begins and the
robot uses a combination of x y positioning code and light recognition code to
move towards the first light. It then
searches for the second light and travels toward the left or right. Then a combination of wall following code
and x y positioning code is used to get FRUSTRATION to the entrance of the
inner U. Once in the U, light
recognition code is used identify where the large trash should be deposited and
IR sensors tell FRUSTRATION when to stop safely. Then the robot exits the U and proceeds to travel through the
remaining coarse. Finally, the robot
passes over the blue and orange storage areas.
Once FRUSTRATION passes the first light, it is continuously searching
for small trash (golf balls) with the camera and will process them according to
color.
Online
Interface:
The final feature of our solution to
the autonomous trash removal problem is a wireless interface that can be used
to inform one of FRUSTRATION’s current status (state, x position, y position,
commanded velocity, etc.) and update key parameters of its code within each
state. This capability enables FRUSTRATION
to keep humans in the loop and ensure steady performance in the face of unknown
disturbances.