💥1 概述
智能导航是移动机器人的关键技术，在移动机器人的应用研究中具有重要地位1。导航主要由构建地图、定位和路径规划三部分构成。.地图构建是指移动机器人根据自身传感器感知周围的环境信息,建立其工作环境模型的过程;定位是指确定移动机器人在其工作环境中的具体位置的过程，即移动机器人在虚拟地图中的坐标和姿态;路径规划是指在复杂的环境中,在满足一定的约束条件下,如行走路径的长度短,路径的代价低、时间短等,找到一条从指定起始位置到指定目标位置的安全、无碰撞的优化路径。对于以上问题,在近几十年的研究中出现了多种有效的解决方案,如:同步建图与地图构建(SLAM)、蒙特卡洛定位算法3动态窗口算法(DWA)等等。
DWA算法是机器人根据传感器获取局部信息的更新,实时控制机器人速度在一个有效的“窗口”内,完成机器人的局部路径规划。
D*算法是基于A*算法基础上提出的,对缺少环境先验信息或者对动态环境下的路径规划解决方案。
📚2 A*算法
A*算法是人工智能中一种典型的启发式搜索算法。通过选择合适的估价函数指导搜索朝着最有希望的方向前进以求得最优解。A*算法针对以往路径规划问题中常常产生的大量无效搜索路径,根据启发函数提高了收索的目的性,提高了算法的搜索效率。A*算法的设计核心是其估价函数。
👨💻3 运行结果
部分代码：
clc
clear
close allmax_x=20;
max_y=20;environment=2*(ones(max_x,max_y));
j=0;
x_val = 1;
y_val = 1;
axis([1 max_x+1 1 max_y+1])
grid on;
hold on;
n=0;xval=18;
yval=18;
x_target=xval;
y_target=yval;environment(xval,yval)=0;%Initialize environment with location of the target
plot(xval+.5,yval+.5,'gd');
text(xval+1,yval+.5,'Target')% obstacle 1
for i = 3:5
for j = 13:16
environment(i,j) = -1;
%plot(i+.5,j+.5,'rd');
end
end
x = [3, 6, 6, 3];
y = [13, 13, 17, 17];fill(x,y,'k');
% obstacle 2
for i = 7:9
for j = 11:14
environment(i,j) = -1;
%plot(i+.5,j+.5,'rd');
end
endx = [7, 10, 10, 7];
y = [11, 11, 15, 15];fill(x,y,'k');
% obstacle 3
for i = 11:13
for j = 11:17
environment(i,j) = -1;
%plot(i+.5,j+.5,'rd');
end
end
x = [11, 14, 14, 11];
y = [11, 11, 18, 18];fill(x,y,'k');
% obstacle 4
for i = 12:13
for j = 6:9
environment(i,j) = -1;
%plot(i+.5,j+.5,'rd');
end
endx = [12, 14, 14, 12];
y = [6, 6, 10, 10];fill(x,y,'k');
% % obstacle 5
% for i = 15:16
% for j = 6:17
% environment(i,j) = -1;
% %plot(i+.5,j+.5,'rd');
% end
% end
%
% x = [15, 17, 17, 15];
% y = [6, 6, 18, 18];
%
% fill(x,y,'k');%plot(19+.5,19+.5,'gd');
xval=3;
yval=3;
x_start=xval;%Starting Position
y_start=yval;%Starting Positionenvironment(xval,yval)=1;
plot(xval+.5,yval+.5,'bo');valid=[]; % x | y | parent_x | parent_y | h(n) | g(n) | f(n)
in_valid=[]; % x | y
k=1;
for i=1:max_x
for j=1:max_y
if(environment(i,j) == -1) % check if obstacle
in_valid(k,1)=i;
in_valid(k,2)=j;
k=k+1;
end
end
end
in_valid_size=size(in_valid,1);cell_x=x_start;
cell_y=y_start;
valid_size=1; % initialize size of valid_list
path_cost=0;
goal_distance=sqrt((cell_x-x_target)^2 + (cell_y-y_target)^2);new_row=[1,8];
new_row(1,1)=1;
new_row(1,2)=cell_x;
new_row(1,3)=cell_y;
new_row(1,4)=cell_x; % parent x
new_row(1,5)=cell_y; % parent y
new_row(1,6)=path_cost; % h
new_row(1,7)=goal_distance; % g
new_row(1,8)=goal_distance; % fvalid(valid_size,:)=new_row; % initializing path with start position
valid(valid_size,1)=0;in_valid_size=in_valid_size+1;
in_valid(in_valid_size,1)=cell_x; % make it invalid for further iterations
in_valid(in_valid_size,2)=cell_y;path_not_found=1;
obstacle_was_detected = 0;% disp('Valid in the beginning');
% valid
% disp('In valid in the beginning');
% in_validwhile 1 % it will be broke only if no new obstacle gets detected just like in A*
while((cell_x ~= x_target || cell_y ~= y_target) && path_not_found == 1)
% x | y | h | g | f
if obstacle_was_detected == 1
fprintf('Value of cell x : %d celly : %d\n',cell_x,cell_y);
end
fprintf('Value of cell x : %d celly : %d\n',cell_x,cell_y);
successors=explore_successors(cell_x,cell_y,path_cost,x_target,y_target,in_valid,max_x,max_y);
successors_size=size(successors,1);
for i=1:successors_size
flag=0;
for j=1:valid_size
if(successors(i,1) == valid(j,2) && successors(i,2) == valid(j,3) ) % if successor same as already existing cell inpath
% disp('valid')
% valid
% disp(' ');
valid(j,8)=min(valid(j,8),successors(i,5)); % check for minimum f and then pick it
if valid(j,8) == successors(i,5)
valid(j,4)=cell_x;% parent x
valid(j,5)=cell_y;% parent y
valid(j,6)=successors(i,3); % h
valid(j,7)=successors(i,4); % g
end;
flag=1;
end;
end;
if flag == 0 % if new cell with minimum f(n) then add to valid_path
valid_size= valid_size+1;
new_row=[1,8];
new_row(1,1)=1;
new_row(1,2)=successors(i,1);
new_row(1,3)=successors(i,2);
new_row(1,4)=cell_x; % parent x
new_row(1,5)=cell_y; % parent y
new_row(1,6)=successors(i,3); % h
new_row(1,7)=successors(i,4); % g
new_row(1,8)=successors(i,5); % f
valid(valid_size,:)= new_row;
end;
end; index_min_cell = min_f(valid,valid_size,x_target,y_target);
if (index_min_cell ~= -1) % if index with minimum fn is obstacle no path exists
cell_x=valid(index_min_cell,2);
cell_y=valid(index_min_cell,3);
path_cost=valid(index_min_cell,6); in_valid_size=in_valid_size+1; % put the cell in_valid so we dont come back on it again
in_valid(in_valid_size,1)=cell_x;
in_valid(in_valid_size,2)=cell_y;
valid(index_min_cell,1)=0;
%valid
%in_valid
else
path_not_found=0;
end;
end;% backtracking to find the path
i=size(in_valid,1);
path=[];
xval=in_valid(i,1); % pick last in in_valid_list that must be target
yval=in_valid(i,2);
i=1;
path(i,1)=xval;
path(i,2)=yval;
i=i+1;if ( (xval == x_target) && (yval == y_target))
% obstacle 5
% tappi = size(in_valid,1) + 1;
% for p = 15:16 %11:13 % w.r.t obstacle 3
% for q = 6:17 %11:17
% environment(p,q) = -1;
% in_valid(tappi,1) = p;
% in_valid(tappi,2) = q;
% tappi = tappi + 1;
% %plot(i+.5,j+.5,'rd');
% end
% end
% in_valid_size= size(in_valid,1);
% x = [15, 17, 17, 15];
% y = [6, 6, 18, 18];
% x = [11, 14, 14, 11]; % with respect to obstacle 3
% y = [11, 11, 18, 18]; fill(x,y,'k');
inode=0;
parent_x=valid(find((valid(:,2) == xval) & (valid(:,3) == yval),1),4);
parent_y=valid(find((valid(:,2) == xval) & (valid(:,3) == yval),1),5);
while( parent_x ~= x_start || parent_y ~= y_start)
path(i,1) = parent_x;
path(i,2) = parent_y;
inode=find((valid(:,2) == parent_x) & (valid(:,3) == parent_y),1);
parent_x=valid(inode,4);
parent_y=valid(inode,5);
i=i+1;
end;
% plottin path
obstacle_was_detected = 0;
j=size(path,1);
p=plot(path(j,1)+.5,path(j,2)+.5,'bo');
%j=j-1;
for i=j:-1:1
pause(.25);
fprintf('Path X : %d Y : %d\n',path(i,1),path(i,2));
% -------------------------check if environment is altered -------------
if environment(path(i,1),path(i,2)) == -1
disp ('obstacle detected');
if(obstacle_was_detected == 0)
% we will start computing the path from the parent path where
% obstacle was first foun
cell_x = valid(find((valid(:,2) == path(i,1)) & (valid(:,3) == path(i,2)),1),4);
cell_y=valid(find((valid(:,2) == path(i,1)) & (valid(:,3) == path(i,2)),1),5);
path_cost=valid(find((valid(:,2) == path(i,1)) & (valid(:,3) == path(i,2)),1),6);
end
obstacle_was_detected = 1;
% remove obstacles from valid list
[valid, valid_size] = remove(valid,path(i,1),path(i,2));
valid = increase_f(valid,path(i,1), path(i,2),in_valid,max_x,max_y);
%disp('Valid after increasing');
%valid
elseif obstacle_was_detected == 0
set(p,'XData',path(i,1)+.5,'YData',path(i,2)+.5);
drawnow ;
else
% fprintf('Removing X : %d Y : %d from invalid\n',path(i,1),path(i,2));
% [x, y] = size(in_valid);
% fprintf('Initial Size : %d x %d\n', x,y); in_valid(find((in_valid(:,1) == path(i,1)) & (in_valid(:,2) == path(i,2)),1),:) = [];
% [x, y] = size(in_valid);
% fprintf('Final Size : %d x %d\n', x,y); valid = increase_f(valid,path(i,1), path(i,2),in_valid,max_x,max_y); % updating costs
end
end
if obstacle_was_detected == 0
plot(path(:,1)+.5,path(:,2)+.5);
break
else
continue
end
else
disp( 'Sorry, No path exists to the Target!');
break
endend
🎉4 参考文献
[1]张希闻,肖本贤.改进D~*算法的移动机器人路径规划[J].传感器与微系统,2018,37(12):52-54+58.DOI:10.13873/J.1000-9787(2018)12-0052-03.
[1]张贺,胡越黎,王权,燕明.基于改进D*算法的移动机器人路径规划[J].工业控制计算机,2016,29(11):73-7477
