93
A School of Robotic Fish for Pollution Detection in Port
1. Avoid obstacles behavior to avoid both stationary and moving
objects based on infrared (IR) sensors. The possible action is selected
from all swim patterns.
2. Follow-wall behavior to keep the robotic fish at a certain distance
from the wall based on the information gathered by IR sensors. The
possible actions are cruise straight, cruise in left/right turn.
3. Wander behavior to explore the tank randomly. No sensor states
are necessary. Possible actions include cruise straight, cruise in left/
right turn, and ascend/descend.
4. Keep-level behavior to keep the robotic fish swimming at a desired
depth level specified by the cognitive layer. It monitors the pressure
sensor data and controls whether the fish ascend or descend.
5. Seek-goal behavior to direct the robotic fish to swim toward the goal;
the possible actions are cruise straight and cruise in right turn.
6. Noise behavior using a random swim pattern to avoid the local minima
and go out from a trap situation by trying all kinds of swim patterns.
7. Feed behavior for a robotic fish to return to a charging station when
its battery is low.
8. Rescue behavior for emergency and self-protection action when
robotic fish face unexpected situations such as motors malfunction,
extreme high internal temperature, etc.
The behavior layer not only decides which swim pattern will be selected
next but adjusts some key parameters for the selected swim pattern. Each
swim pattern has tens of parameters in its function to specify its performance. However, to simplify the control loop, the behavior layer only adjusts
two parameters. More details can be found in Liu and Hu (2010).
5.4.3 Swim Pattern Layer
Our project aims to design and build an autonomous navigation robotic
fish that would swim like a real fish and realize autonomous navigation.
To realize the fish-like motion, we have designed three or four tail joints for
our robotic fish. The added-mass hydrodynamic theory is adopted here to
achieve seven fish-like swim patterns as follows.
1. Cruise straight: The fish swims along a straight line at a constant
speed, possibly with small acceleration/deceleration.
2. Cruise in turning: Fish is turning in a small angle and at a constant
speed.
A School of Robotic Fish for Pollution Detection in Port
1. Avoid obstacles behavior to avoid both stationary and moving
objects based on infrared (IR) sensors. The possible action is selected
from all swim patterns.
2. Follow-wall behavior to keep the robotic fish at a certain distance
from the wall based on the information gathered by IR sensors. The
possible actions are cruise straight, cruise in left/right turn.
3. Wander behavior to explore the tank randomly. No sensor states
are necessary. Possible actions include cruise straight, cruise in left/
right turn, and ascend/descend.
4. Keep-level behavior to keep the robotic fish swimming at a desired
depth level specified by the cognitive layer. It monitors the pressure
sensor data and controls whether the fish ascend or descend.
5. Seek-goal behavior to direct the robotic fish to swim toward the goal;
the possible actions are cruise straight and cruise in right turn.
6. Noise behavior using a random swim pattern to avoid the local minima
and go out from a trap situation by trying all kinds of swim patterns.
7. Feed behavior for a robotic fish to return to a charging station when
its battery is low.
8. Rescue behavior for emergency and self-protection action when
robotic fish face unexpected situations such as motors malfunction,
extreme high internal temperature, etc.
The behavior layer not only decides which swim pattern will be selected
next but adjusts some key parameters for the selected swim pattern. Each
swim pattern has tens of parameters in its function to specify its performance. However, to simplify the control loop, the behavior layer only adjusts
two parameters. More details can be found in Liu and Hu (2010).
5.4.3 Swim Pattern Layer
Our project aims to design and build an autonomous navigation robotic
fish that would swim like a real fish and realize autonomous navigation.
To realize the fish-like motion, we have designed three or four tail joints for
our robotic fish. The added-mass hydrodynamic theory is adopted here to
achieve seven fish-like swim patterns as follows.
1. Cruise straight: The fish swims along a straight line at a constant
speed, possibly with small acceleration/deceleration.
2. Cruise in turning: Fish is turning in a small angle and at a constant
speed.