Biomedical Engineering Reference
In-Depth Information
&
An ultrasound distance sensor for obstacle avoidance
&
A compass sensor for heading control
&
Contact/pressure sensors at foot tips
It is important to note that the legs themselves can be used as exteroceptive
sensors. One can use the current sensors of the joint motors as a tactile sensor
during movement, e.g., whether a leg presses against an object can be sensed by
an increasing current in the corresponding joint motor. This issue is currently
under further investigation. In an ongoing project, we are analyzing the robot's
capabilities to use its front legs in order to determine shape information of ob-
stacles, which it previously detected with ultrasound sensors.
In order to allow an operator to communicate with the robot or to take data
samples during a test run, the robot is equipped with a wireless 28K-baud
bidirectional communication link and a PAL CCD camera with a 5-GHz video/
audio link for video transmission. It is thus possible to use the robot as a
semiautonomous system. The operator can control it via high-level commands
like "walk forward," "left," "right," "go up," "go down," "move sideways," and
"turn." To supervise the system, all relevant sensor data are sent back from the
robot to the operator.
2.2. The Processing Hardware
A network of Infineon C167 and C164 microcontroller derivates are used,
containing one master controller (the C167), which functionally contains the
higher behavioral level, the communication to the operator, and executes proc-
essing of data from the exteroceptive sensors. The master controller is connected
via a CAN-bus network with leg controllers (the C164). The leg controller exe-
cutes local controller functions (via the central pattern generator (CPG)) and
local reflex control functions, processes the proprioceptive sensor data, and con-
trols the DC motors. In a new version of the hardware we will use an Motorola
MPC555 master controller for higher-level control and an FPGA for local con-
trol of the legs.
3.
AMBULATION CONTROL
Our architecture (14) is based on two approaches to robust and flexible real-
world locomotion in biological systems. These are the central pattern generator
(CPG) model and the coupled reflex approach (1,6,7).
A CPG is able to produce a rhythmic motor pattern even in the complete
absence of sensory feedback. The general model of a CPG has been identified in
