Brian's ROACH Page
RObotic Automation Control Hierarchy
Here you can see the white camera, and conical lense hood of the IR
pyroelectric sensor on the pan/tilt head. The pan (stepper)/tilt (servo)
mechanism is partially obscured.
You can see the main power batteries arranged in a triangular configuration.
The SRAM backup power is supplied by 4x1.2VDC NiCd batteries located in
the center of the triangle. Video system power is supplied by a 12 volt
battery pack on the underside of the robot.
At the rear of the robot are two BNC connectors. The one closest to
the camera is for the video transmitter (not mounted yet), and the antenna
mounted farther away from the camera is for the radio modem (currently
mounted under the upper shell).
The ROACH is rapidly nearing completion. All major structural areas
have been finished. The locomotion, and pan/tilt interface are tested and
working. The shaft of the stepper motor is directly bound to the sensor
array which greatly limits phase stepping strength. This approach also
adds an annoying overswing. I am considering replacing the stepper motor
with a gear reduced DC motor, or stepper motor.
Check out the ROACH's Software
1-1-98 The ROACH project has
been completed. The information generated from ROACH research has allowed
me to move on to bigger and more complex robotic systems such as the Mowbot,
and Blue Cube. In
retrospect, the robot base could have been better designed, and use lighter
batteries to reduce motor current draw. The IR sensors used worked far
better than expected, and will be considered for use in further projects
(thanks MIL!) the 68HC11 is a great
low-end micro, and is an excellent teaching tool for embedded systems when
power conservation, and size are the most important factors. PS. The Proxim
Proxlink board is still available from Timeline
4-4-97 I was having a few problems with the servo->sensor array
linkage. The joint didn't appear to be strong enough to support the full
weight of the sensor array, but I fixed it! I hacked apart an old printer
to find a very specific part. It originally housed two gears that meshed
together within a metal 'U'. Three hours of cutting and machining with
my Dremel tool and it was in place. I can finally continue mounting the
sensor array! I have decided to mount the sensor array to the pan-tilt
mechanism with 'carpet tape' for safety reasons. After all, I don't want
the entire thing to snap off if it hits something.
Do power/weight/component AH calculations
Permanently mount motor power cells to base.
Design, build docking station, connector mating system.
Purchase Dinsmore 1490 digital compass, mount on 6.270 board.
Here you can clearly see the old weak servo/stepper link, and the secure,
yet kludgely looking support fixture. It only looks bad, it's really quite
Favorite Related Links
Another robot named ROACH.
A real Robo-Roach!
Some confrencee's experiences/code for AAAI
'93, pay special notice to the Subsumption code in Jeffrey Graham's
MIT's Cherupakha FTP site.
NASA's Inteligent Mechanism
MONDOTronics web page
Brandeis University's Maja Mataric (formerly of MIT's Insect Lab) now has
her own Robotic Interaction
A very informative Cognitive/Agent
Architecture page at U-Mich.
A really cool Polypod
robot locomation implimentation.
Tons of info on Ultrasonic
Last update 3-16-1998