|
Steering and Schemas
Behavior-based robotic control merges simple reactive "behaviors" into a
more complex resulting behavior. This can be useful if you want a robot to say,
move towards a waypoint while avoiding obstacles,
while seeking a certain color, all at the same time. One problem
that can be encountered by this approach is for a robot
to get stuck in a cyclical pattern where the
combination of behaviors driving it do not provide a way forward.
The clip
on the left is from a simulation
of a robot (blue, bottom-left,) running goal-seeking, collision-avoidance,
and blob-tracking behaviors concurrently.
Its task is to travel from one green cone to the other and back again.
First, only the combination of behaviors is used.
Second, a path planner is added in order to plan a route past a "no-go-zone".
By using a little planning, behaviors can
be coordinated or given inputs that will result in more successful outputs.
Another approach is to add a behavior whose task it is to detect the
stall condition and actively move away from it, with no need to deliberate on why it is
occuring.
simulation run in
Player/Stage
|
