|
||||||||
Robot
75 - Jeeves
|
The Objective: The
objective of the contest is for the robot to navigate through the course to 5
pre-designated waypoints. When traversing the course, the robot must avoid
obstacles (2’x2’ square boxes) and collect colored golf balls along the way.
For a successful run, the robot will collect & categorize at least two
golf balls, and deploy them into the ball chutes after reaching all
waypoints. |
||
Path Planning: The
robot primarily utilizes the A-star path planning algorithm in order to move
about the course. The premise of this algorithm is to build an optimal path
from the robot’s location to each waypoint, using intermediate waypoints. The
algorithm check through a list of “available nodes”, or locations on the map
that do not have obstacles, and then uses heuristics to find an optimal path
to the waypoint. In the later stages of the project the team implemented
diagonal A-star which allowed the robot to travel on a diagonal and not be
tied to a grid system. This allowed the robot to run more efficient paths. |
||
Ball Collection: A
key component of the contest is ball collection and deployment. Our robot
utilizes image processing for golf ball detection. A camera, mounted on the
front of the robot, processes video and thresholds the color values being
sent through. Blue and orange color values are marked as golf balls after
additional thresholding. Once found, the robot centers the golf ball in the
camera, before running through an open loop collection method. Orange and
blue golf balls are contained in different compartments in the robot’s
“gripper”. Each compartment has its own gate to ensure the golf balls remain
in the gripper until deployment. |
||
LabVIEW: The
LabVIEW interface is a useful tool for visualizing what the robot encounters when
traversing the course. The LabVIEW environment collects and plots the
obstacles that the robot sees in the course, the golf balls it collects, and
keeps track of the robot’s path along the way. |
||
Obstacle
Avoidance: Before
traversing the course, the only prior information that the robot know are the
external course dimensions and the location of the waypoints the robot must
travel to. The robot must be able to locate and avoid up to five, 2’x2’
square obstacles. The robot uses a LADAR (Light Detection and Ranging) sensor
to range solid objects in the course. Whenever a LADAR reading is received by
the robot, its value is compared to the values for known possible locations
of obstacles. If each obstacle is positively identified 5 times, the robot
will avoid those tiles when running its path planning algorithm. |
||
|