|
HaViMo
HaViMo stands for
Hamid’s Vision
Module, named after its inventor. It is a module that give Bioloid
the ability to navigate in a familiar environment such as a Robocup
football field or a maze. The module identifies up to 16
contiguous color areas and maintains, for each such area, many
parameters including: Color ID, Average coordinate, Bounding box,
#pixels, etc.
- The included PC program
allows users to
define sets of discrete RGB colors (via 3D color sets). The color sets
are used
by the module in search of contiguous areas in captured images. It
allows for adjustment
of camera settings as well as color definitions in the look-up table.
All settings are be saved into the FLASH/EEPROM memory of the module.
This means that under the same lighting and color conditions, there
is no need to calibrate the device after every power on. After
calibration, the module is ready to be used in implementation phase.
In this phase the module is only connected to the CM5 and can receive
commands to sample and process the image and return image processing
results.
- HaViMo maintains a large
table with
about 50 parameters that some can be read and others can be read and
written to. Using the Bioloid AX12 Dynamixel serial protocol, the bus
controller can ask the HaViMo 8 times/sec to send back information
regarding the state of the 16 contiguous-color-areas (stored
logically in this addressable table).
- It lets a relatively low
end
processor such as an Atmega128 to receive simple digests of image
interpretation and in turn direct the robot towards targets such as a
ball or our of a labyrinth.
HaViMo was developed by
Berlin's Free
University for the RoboCup humanoid robots of FU-Berlin, the
“FUmanoids”, that
won the 3 Place of the RoboCup 2007 Humanoid League in Atlanta.
Features:
- Lightweight integrated camera and image
processor
- Embedded into a Bioloid AX12 servo-box
- Controled
and Queried via Bioloid
Dynamixel protocol
- Able to Acquire images and send
them to
a remote PC
- Recognizes objects in real-time using
online image recognition techniques
- Online patern
recognition allows simple
microcontroller to track objects and self localize.
Technical Details:
- Frame rate: 8 FPS
- Resolution:
160*120 color
- Built in color lookup table
- Image
noise filtering
- Region growing for up to 255
different
color categories
- Detection of up to 16 contiguous
color
regions at 8hz in the captured Image
- RS485 (Bioloid
Dynamixel) interface,
can be connected to any type of microcontroller having USART or
directly to the Bioloid RS485 Bus.
- Output Format:
256 bytes after
processing of each frame, containing information about the found
color regions on the image, including coordinates, dimensions,...
results can be easily used by CM5 programmed using C.
- Communication
protocol: Bioloid
standard communication protocol
- Included PC
software: 3D Color category
definition, Picture Sample, Access to the camera settings.
See also:
|