ABSTRACT
Recently, multiple unmanned ground vehicles (multi-UGVs) have attracted a great deal of attention as viable solutions to a wide variety of military and civilian applications. Among many topics in the field of multi-UGVs, the formation cooperative reconnaissance is of great importance, which helps to concentrate firepower to suppress and eliminate the suspicious enemies. This paper presents a formation cooperative reconnaissance strategy for multi-UGVs to explore all the unknown areas in the environment, which is similar to the human soldiers paraded on patrol. First, the frontier-based exploration algorithm is proposed to get the reconnaissance targets in the unknown areas which is modeled as a global grid map. Then, A*-based multi-target path planning is applied to find the right exploring order of reconnaissance targets. Lastly, the distance-angle leader-follower formation control laws are designed to make the multi-UGVs rapidly form a desired formation after assigning a reconnaissance goal. The experiment results show that all the reconnaissance targets are searched and explored. The follower UGVs are gradually adjusting their position and orientation to follow the leader UGV while moving to each reconnaissance target sequentially. The motion of the multi-UGVs is reliable. Especially, the average position and heading errors of formation are less than 0.2%.
CO EDITOR-IN-CHIEF:
ASSOCIATE EDITOR:
Nomenclature
C | = | auxiliary variable |
dij | = | distance from the center of mass of jth to ith unmanned ground vehicles |
ei | = | ith frontier region |
Ef | = | set of frontier regions |
E | = | set of directed edges in the graph |
Ed(ei, ej) | = | the Euclidean distance between frontier region ei and ej |
g(s) | = | cost between start state sstart and state s |
G | = | conversion matrix |
h(s) | = | heuristic estimate cost between state s and goal state sgoal |
k1 and k2 | = | controller gain |
M(s) | = | upper right submatrix |
N | = | zero mean Gaussian white noises with covariance PN |
N(s) | = | upper left submatrix |
O | = | zero mean Gaussian white noises with covariance Pl |
S | = | set of vertices in a discretized finite state-space |
si | = | ith state in the finite state-space |
sstart | = | start state |
sgoal | = | goal state |
scurr | = | current state |
s | = | state vector of tracking error |
u | = | control vector |
ui | = | control vector of ith unmanned ground vehicle |
uj | = | control vector of jth unmanned ground vehicle |
UGV_n | = | nth unmanned ground vehicle |
v | = | translational velocity |
x | = | coordinate along the x-axis |
xi | = | x-coordinate of ith unmanned ground vehicle |
xj | = | x-coordinate of jth unmanned ground vehicle |
y | = | coordinate along the y-axis |
yi | = | y-coordinate of ith unmanned ground vehicle |
yj | = | y-coordinate of jth unmanned ground vehicle |
y | = | output observation vector |
z | = | state vector |
αij | = | heading angle difference between jth and ith unmanned ground vehicles |
δij | = | angle from the y-axis of ith leader unmanned ground vehicle to dij |
Δi | = | heading angle deviation between 0th and ith unmanned ground vehicles |
θ | = | heading angle |
θi | = | heading angle of ith unmanned ground vehicle |
θj | = | heading angle of jth unmanned ground vehicle |
πi | = | ith returned optimal path with start state sstart and goal state sgoal |
Πi | = | returned optimal path with start state sstart and goal state ei |
φij | = | angle from the y-axis of jth follower unmanned ground vehicle to dij |
ω | = | angular velocity |
Disclosure statement
No potential conflict of interest was reported by the author(s).