Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
MASLAB'S Autonomous
thresholds determined by trial and error. Besides the giant wheels, the Maciej and Huan
watch their
Although this worked, the team worried team made their robot, nicknamed
robot navigate
that it would not allow for differences in Arnold, distinct with an unusual
S
ixty MIT students spent SIMPLE SOLUTIONS LEAD TO
with knowledge of its own shape, this plates were controlled by a belt and
January building autonomous A 2ND-PLACE STANDING
enabled the robot to navigate without any guided by rods that the team har-
vision-based robots as part of As an all-freshman team,
unexpected wall collisions. Even though vested from a used inkjet printer. To
the Mobile Autonomous Stephanie Chin (Ocean
their camera-based navigation system prevent this mechanism from bind-
Systems Laboratory (Maslab). The object Engineering), Sarang Kulkarni,
worked, the team added IR sensors for extra ing, they relied on a heavy coat of WD-40.
was to build robots that could navigate and Andrew Lewine (Electrical
security during the final competition. The giant wheels and complicated scoring
an unknown playing field to find balls Engineering and Computer
mechanism made designing and program-
and put them in goals. Each team of two Science) felt they to make a simple
GIANT WHEELS ming this robot difficult. In the end, this
to five students started with basic materi- robot. Their approach was to
When Ben Switala (EECS), Jason caused the robot to not be entirely
als: a square piece of pegboard, two rub- make a circular robot with a front
Bryslawskyj (Physics), Ted Blackman (EECS debugged. Neverthe-less, it was a favorite
ber wheels with drive motors, a caster, a gate that would open as the robot
and Physics), and Melodie Kao among those watching the final competi-
small computer programmable in Java, a drove forward to collect balls.
(Architecture) saw a pair of 9-inch-diameter tion, Maslab students, and the Maslab staff
Road Narrows Robotics ORC controller Once in front of a goal, the robot
wheels, they were inspired to add them to for its brazen design.
board, and a color web camera. The stu- would open its gate, drive back-
their robot. These wheels became the most
dents expanded on these building blocks ward to release its balls, close its
defining aspect of their robot, as they made THREE STRATEGIES
with their choice of materials to create gate, and drive forward to push
it both incredibly large and unusually fast. OF WANDERING
custom robots. They were given an the balls into the goal. When not
Mark Stevens (EECS), Jeremy
opportunity to test their designs at the collecting balls or scoring goals,
Smith (EECS), Alex St. Claire
end of January in a final competition in the robot explored the field using
(Mechanical Engineering), and
which robots scored points by taking short-range IR sensors to follow
Jenny Liu (EECS) built a robot
possession of balls, putting them into the wall. A robot navigating the final playing field toward a red
with effective software and a
goals, and putting them in front of goals. The team opted for simple ball.
solid mechanical design. Their
Students were not told the configuration solutions to problems that other teams
NEURAL NETWORK SEES strategy was to design their
of the final competition playing field spent a great deal of time overcoming.
Omobayonle Olatunji (EECS), Huan Liu robot's hardware to simplify its Daniel talking to staff member Timothy Abbott
beforehand, but they did know some For example, instead of writing software
(EECS), Maciej Pacula (EECS), Matthew code. For example, the robot's about starting up his team's robot during the final
basics. They knew that the playing field to filter out camera data above the top of
ball-capture mechanism con- competition.
was surrounded by white walls topped the playing field walls, they angled their Gordon (Math), and Alexandre Oliveira
created a robot that used a strategy simi- sisted of a rotating drum that
with a thick blue line and periodic blue camera so that it could not see above the
would push balls up a ramp ment requirements for the software. The
tick marks. Goals consisted of rectangu- walls. With the time they saved building lar to that of the freshman team's robot,
and into an internal storage team also made it a priority to construct
lar holes cut into the wall surrounded by a simple robot, the team was able to rig- but relied on no sensors except its cam-
bin. The mouth of this mecha- their robot's physical mechanisms rapidly.
a thick yellow line. Strips of green and orously test their design. They brain- era.
nism was wide, reducing align- This allowed software to be tested on a
black squares, called barcodes, were put stormed potentially problematic situa- To do this, they needed to differenti-
working robot early, and pro-
on the walls so they were connected by tions for their robot, made appropriate ate colors very reliably. Data from the
vided more opportunities for
line of sight. These were intended to help changes to the software, and tested the robot's camera was accessible to its com-
discussion between software
robots navigate the field. Finally, the robot on mock playing fields to expose puter in the form of numbers represent-
and mechanical team members.
floor was made of black carpet, and balls further vulnerabili- ing the color and brightness of each cam-
The team developed a num-
were painted bright ties. In the rush to era pixel. To decide which numbers rep-
ber of wandering strategies for
red. create a robot, very resented which colors, other students set
exploring the playing field.
Each team pro- few teams found the Unfortunately, no single strate-
duced a unique robot time to improve their gy worked perfectly. To deal
with its own design by testing with this, the team decided to
strengths and weak- their robot and use three different wandering
nesses. The following uncovering its weak- algorithms and alternate
descriptions high- nesses. By keeping between them randomly. These
light just a handful of things simple, this approaches were to travel along
the teams and their team created a robot Left: A picture used by the team relying only the nearest wall, drive toward
creations for the 2007 capable of a second- on vision to test their neural network for red the farthest open distance
Maslab competition. Andrew, Stephanie and Sarang working place ranking in the color identification. Right: The result of this The giant-wheeled robot with team member Melodie. Mark, Jeremy, Alex and Jenny designed this “wan-
detected with an IR sensor, and
on their simple round robot's code. identification dering” robot.
final competition.
80 R O B O T M A G A Z I N E SUMMER 2007 81
E D U B O T S
82 R O B O T M A G A Z I N E