A Novel Distributed Hybrid Cognitive Strategy for Odor Source Location in Turbulent and Sparse Environment

View full-text article in PMC

. 2025 Aug 4;27(8):826. doi: 10.3390/e27080826

Algorithm 1 CGRInfotaxis for Multi-Agent System

Input: Searching environment parameters, particle filter parameters, and agent parameters.

Output: Action of multi-agent and the position of predicted source

1:
for $k = 1, 2, \dots k_{m a x}$ do
2:
for all agent $n \in {1, 2, \dots, N}$ do
3:
Observation: $z_{k, n}$ read multi-agent measurements;
4:
Share Sensor Measurements: Broadcast $z_{k, n}$ to all agents and receive $z_{k} = {z_{k, 1}, z_{k, 2}, \dots, z_{k, N}}$ ;
5:
end for
6:
$w_{(k, n)}^{(m)} \leftarrow w_{(k - 1, n)}^{(m)}$ using Equations (30) and (31);
7:
if $M_{e f f} < M_{t h}$ then
8:
Resampling and set $w_{k}^{(m)} = \frac{1}{M}$ ;
9:
end if
10:
Compute Initial Temporary Decisions for all Agents
11:
while not Decisions reach consensus do
12:
for all agent $n \in {1, 2, \dots, N}$ do
13:
Decision Sharing: Broadcast temporary decision and receive other agent decisions
14:
for all $a_{k} \in A_{k}$ do
15:
$p ({\hat{z}}_{k + 1} | θ_{k + 1})$ for other agent decisions and $a_{k}$ using Equation (29)
16:
$w_{(k + 1, n)}^{(m)}$ using Equation (30);
17:
$I_{R \overset{´}{e} n y i_} (a_{c k})$ computation using Equation (25);
18:
$G (a_{c k})$ computation using Equation (26);
19:
$J (a_{c k})$ computation using Equation (33);
20:
end for
21:
$a_{(c k, n)}^{*} = max (J (a_{c k})) = max [(1 - e^{- λ k}) * G_{c o o} + e^{- λ k} * I_{r \overset{´}{e} n y i_}]$ select the motion control;
22:
Update the decision of agent n
23:
end for
24:
Check consensus status:
25:
if Decisions reach consensus then
26:
output the action of multi-agent
27:
$[x_{k + 1}, y_{k + 1}]$ of every agent computation using Equation (14)
28:
end if
29:
end while
30:
if source confirmation conditions reached then
31:
output the position of predicted source
32:
break
33:
end if
34:
end for