CN103116356B - Method of search in mazes - Google Patents

Abstract

The invention relates to a method of search in mazes. The method includes the steps of firstly, generating a non-directional probabilistic distance map; secondly, generating a directional probabilistic distance map; and thirdly, applying the probabilistic distance maps. After the directional probabilistic distance map in the step 2 is generated, a robot starts from a starting maze cell to search a maze. When encountering an intersection, the robot judges to which one of the maze cells in certain directions the minimum probabilistic distance belongs so as to determine an advancing path by means of the probabilistic distance maps according to the current direction, and reaches a target area G. 'Probabilistic distance' is provided by organically combining probability in the science of probability and distance in kinematics. Based on the probabilistic distance, search in mazes is rational and rules-based. The method of search in mazes based on 'probabilistic distance' is realized.

Description

The method of a kind of labyrinth search

Technical field

The present invention relates to the method for a kind of labyrinth search, belong to the technical field of artificial intelligence and robot.

Technical background

Along with the development in epoch, artificial intelligence and Robotics more and more very powerful and exceedingly arrogant, study and research tide just have swept the globe.The research that associated machine people walks labyrinth is also in the ascendant, and various algorithm emerges in an endless stream, and has greatly promoted the development of artificial intelligence and Robotics.

How to keep away barrier walk out smoothly in the process in labyrinth in research robot, adopt which kind of method search, solution efficiency that traverse maze directly affects maze problem, be the marrow place solving maze problem.Current conventional searching method is divided into two classes, and a class is non-Graph-theoretical Approach, as left hand, right hand method; Another kind of is Graph-theoretical Approach, as depth-first method, and centripetal method etc.The general efficiency of these methods is not high, robot can not be made to pass through labyrinth fast and reach the destination.

In sum, how to search for for labyrinth, propose a kind of high efficiency searching method, fast and effeciently solve the problem of passing through arrival destination, labyrinth.The present invention can be widely used in the search of various types of labyrinth, also can be used for reference by the application of other field.

Summary of the invention

For the deficiencies in the prior art, the invention discloses the method for a kind of labyrinth search.

Technical scheme of the present invention is as follows:

A method for labyrinth search, comprises step as follows:

1) directionless probability metrics figure is generated:

In labyrinth, target area G is four maze lattices communicated at center, labyrinth, this target area with extraneous adjacent be eight maze lattices, target area must communicate with the external world, and dotted line expresses possibility the wall existed, and the grid that dotted line surrounds is maze lattice;

When supposing initial, other walls all do not exist, the spacing of maze lattice is designated as 1, and each 90 degree of turning spent time is also equivalent to distance, is designated as a, each is turned through 180 degree spent time and is equivalent to distance b (robot a ≈ 1, the b ≈ 10 used in test).

Position residing for robot is maze lattice A, then the computing method of the distance of maze lattice A and target area G are: the wall non-existent probability nearest apart from maze lattice A is p ₁=2 ⁷/ (2 ⁸-1), now the bee-line of maze lattice A to target area G is s ₁, namely maze lattice A has p to target area G ₁potential range be s ₁; Deposit in case at described nearest wall, the non-existent Probability p of wall that distance maze lattice A second is near ₂=(1-p ₁) 2 ⁶/ (2 ⁷-1), now the bee-line of maze lattice A to target area G is s ₂, namely maze lattice A has p to target area G ₂possible bee-line be s ₂; The rest may be inferred, p _n, s _nas shown in table 1:

Table 1:

Finally can obtain the distance of maze lattice A to target area G as (1) formula:

S _A＝p ₁s ₁+p ₂s ₂+p ₃s ₃+p ₄s ₄+p ₅s ₅+p ₆s ₆+p ₇s ₇+p ₈s ₈(1)

Due to the product that the above results is a probability and distance, represent the distance on probability meaning, the S in formula (1) _abe called probability metrics;

2) generation has Direction Probability distance map:

In step 1) described probability metrics S _acomputation process in, do not consider the direction of robot current kinetic, this is obviously irrational, because each turning or tune, capital affects the probability metrics between robot and target area G, so the current direction of robot is also one of factor affecting probability metrics.

As robot initial working direction upwards, then maze lattice B has p to target area G ₁possible bee-line be s _1dir; There is p ₂possible bee-line be s _2dir; The like can obtain p _n, s _ndiras shown in table 2 below:

Table 2:

The oriented probability metrics of maze lattice B to target area G can be obtained, shown in (2):

$\begin{array}{c}{S}_{\mathrm{Bdir}}=p_1{s}_{1\mathrm{dir}}+p_2{s}_{2\mathrm{dir}}+p_3{s}_{3\mathrm{dir}}+p_4{s}_{4\mathrm{dir}}+p_5{s}_{5\mathrm{dir}}+p_6{s}_{6\mathrm{dir}}+p_7{s}_{7\mathrm{dir}}+p_8{s}_{8\mathrm{dir}}\\ =2\frac{209}{255}+2\frac{134}{255}a\\ \≈5\frac{88}{255}\end{array}---\left(2\right)$

Coordinate according to labyrinth cell is numbered, and calculates according to above-mentioned algorithm, show that in labyrinth, each cell is to the probability metrics of target area:

When robot initial direction is upper, obtain the probability metrics figure in the labyrinth of " upward direction " according to above-mentioned calculating and statistical statistics;

In like manner, when robot initial direction is right, the probability metrics figure in the labyrinth of " upward direction " is turned clockwise 90 ° (keeping coordinate constant, only numerical value variation);

Robot initial direction be under time, the probability metrics figure in the labyrinth of " upward direction " is turned clockwise 180 ° identical (keep coordinate constant, only numerical value variation);

When robot direction is left, the probability metrics figure in the labyrinth of " upward direction " is rotated counterclockwise 90 ° identical (keeping coordinate constant, only numerical value variation).

3) application of probability metrics figure

Establishment step 2) described in have Direction Probability distance map after, robot is from starting point maze lattice search labyrinth, crossing is run in advancing, according to the current direction of robot, utilize probability metrics figure, judge that the probability metrics value of the maze lattice distance objective in which direction is minimum, decide course with this, until robot arrives at target area G.

Advantage of the present invention and good effect are:

1, the present invention is applicable to the search in various labyrinth, for robot path choice provides theoretical foundation, has filled up blank in this regard.

2, invention increases the efficiency of labyrinth search, shorten labyrinth search time, and be very easy to dispose, implement.

3, the present invention is novel unique, for thinking has been opened up in the further research of artificial intelligence and Robotics.

4, the probability in Probability organically combines with the distance in kinematics by the present invention, proposes " probability metrics " this parameter, with this parameter for benchmark, make labyrinth search for reasonable can according to, have regulations to abide by.Present invention achieves the maze search method based on " probability metrics ".

Accompanying drawing explanation

Fig. 1 is the labyrinth example of embodiment, and middle four cells are target area G;

Fig. 2 be robot initial direction upwards time corresponding probability metrics figure.

Embodiment

Below in conjunction with embodiment and Figure of description, the present invention is described in detail, but is not limited thereto.

Embodiment

1) directionless probability metrics figure is generated

In labyrinth, target area G is four maze lattices communicated at center, labyrinth, what G same external world in this target area was adjacent is eight maze lattices, target area G must communicate with the external world, also be have seven wall walls at the most around the G of target area, as shown in fig. 1, dotted line express possibility exist wall, the grid that dotted line surrounds is maze lattice.

When supposing initial, other walls all do not exist, the spacing of maze lattice is designated as 1, and each 90 degree of turning spent time is also equivalent to distance, is designated as a, each is turned through 180 degree spent time and is equivalent to distance b (robot a ≈ 1, the b ≈ 10 used in test).

In Fig. 1, the distance of maze lattice A and target area G should calculate like this: the wall non-existent probability nearest apart from maze lattice A is p ₁=2 ⁷/ (2 ⁸-1), now the bee-line of maze lattice A to target area G is s ₁, namely maze lattice A has p to target area G ₁potential range be s ₁; Deposit in case at this wall, the non-existent Probability p of wall that distance maze lattice A second is near ₂=(1-p ₁) 2 ⁶/ (2 ⁷-1), now the bee-line of maze lattice A to target area G is s ₂, namely maze lattice A has p to target area G ₂possible bee-line be s ₂; The rest may be inferred, p _n, s _nas shown in table 1 below:

Table 1:

Finally can obtain the distance of maze lattice A to target area G as shown in the formula (1):

S _A＝p ₁s ₁+p ₂s ₂+p ₃s ₃+p ₄s ₄+p ₅s ₅+p ₆s ₆+p ₇s ₇+p ₈s ₈(1)

Due to the product that the above results is a probability and distance, represent the distance on probability meaning, the S in formula (1) _afor probability metrics;

2) generation has Direction Probability distance map

For Fig. 1, suppose robot initial working direction upwards, maze lattice B has p to target area G ₁possible bee-line be s _1dir; There is p ₂possible bee-line be s _2dir; The like can obtain p _n, s _ndiras shown in table 2 below:

Table 2:

The oriented probability metrics of maze lattice B to target area G can be obtained such as formula (2):

$\begin{array}{c}{S}_{\mathrm{Bdir}}=p_1{s}_{1\mathrm{dir}}+p_2{s}_{2\mathrm{dir}}+p_3{s}_{3\mathrm{dir}}+p_4{s}_{4\mathrm{dir}}+p_5{s}_{5\mathrm{dir}}+p_6{s}_{6\mathrm{dir}}+p_7{s}_{7\mathrm{dir}}+p_8{s}_{8\mathrm{dir}}\\ =2\frac{209}{255}+2\frac{134}{255}a\\ \≈5\frac{88}{255}\end{array}---\left(2\right)$

For IEEE computer mouse contest standard labyrinth used, the coordinate according to labyrinth cell is numbered.When robot initial direction is upper, obtain the probability metrics figure in the labyrinth of " upward direction " according to above-mentioned calculating and statistical statistics, as shown in Figure 2;

In like manner, when robot initial direction is right, the probability metrics figure in the labyrinth of " upward direction " is turned clockwise 90 ° (keeping coordinate constant, only numerical value variation);

Robot initial direction be under time, the probability metrics figure in the labyrinth of " upward direction " is turned clockwise 180 ° identical (keep coordinate constant, only numerical value variation);

When robot direction is left, the probability metrics figure in the labyrinth of " upward direction " is rotated counterclockwise 90 ° identical (keeping coordinate constant, only numerical value variation).

3) application of probability metrics figure

Establishment step 2) described in have Direction Probability distance map after, robot is from starting point maze lattice search labyrinth, crossing is run in advancing, according to the current direction of robot, utilize probability metrics figure, judge that the probability metrics value of the maze lattice distance objective in which direction is minimum, decide course with this, until robot arrives at target area G.

Claims (1)

1. a method for labyrinth search, it is characterized in that, it is as follows that the method comprising the steps of:

1) directionless probability metrics figure is generated:

In labyrinth, target area G is four maze lattices communicated at center, labyrinth, this target area G with extraneous adjacent be eight maze lattices, target area G must communicate with the external world, and dotted line expresses possibility the wall existed, and the grid that dotted line surrounds is maze lattice;

When supposing initial, other walls all do not exist, and the spacing of maze lattice is designated as 1, and each 90 degree of turning spent time is also equivalent to distance, is designated as a, and each is turned through 180 degree spent time and is equivalent to distance b, robot a ≈ 1, the b ≈ 10 used in test;

Position residing for robot is maze lattice A, then the computing method of the distance of maze lattice A and target area G are: the wall non-existent probability nearest apart from maze lattice A is p ₁=2 ⁷/ (2 ⁸-1), now the bee-line of maze lattice A to target area G is s ₁, namely maze lattice A has p to target area G ₁potential range be s ₁; Deposit in case at described nearest wall, the non-existent Probability p of wall that distance maze lattice A second is near ₂=(1-p ₁) 2 ⁶/ (2 ⁷-1), now the bee-line of maze lattice A to target area G is s ₂, namely maze lattice A has p to target area G ₂possible bee-line be s ₂; The rest may be inferred, p _n, s _n;

Finally can obtain the distance of maze lattice A to target area G as (1) formula:

S _A＝p ₁s ₁+p ₂s ₂+p ₃s ₃+p ₄s ₄+p ₅s ₅+p ₆s ₆+p ₇s ₇+p ₈s ₈(1)

Due to the product that the above results is a probability and distance, represent the distance on probability meaning, the S in formula (1) _afor probability metrics;

2) generation has Direction Probability distance map:

As robot initial working direction upwards, then maze lattice B has p to target area G ₁possible bee-line be s _1dir; There is p ₂possible bee-line be s _2dir; The like can obtain p _n, s _ndir; The oriented probability metrics of maze lattice B to target area G can be obtained, shown in (2):

$\begin{array}{c}{S}_{\mathrm{Bdir}}=p_1{s}_{1\mathrm{dir}}+p_2{s}_{2\mathrm{dir}}+p_3{s}_{3\mathrm{dir}}+p_4{s}_{s4\mathrm{dir}}+p_5{s}_{5\mathrm{dir}}+p_6{s}_{6\mathrm{dir}}+p_7{s}_{7\mathrm{dir}}+p_8{s}_{8\mathrm{dir}}\\ =2\frac{209}{255}+2\frac{134}{255}a\\ \≈5\frac{88}{255}\end{array}---\left(2\right)$

Coordinate according to labyrinth cell is numbered, and calculates according to above-mentioned algorithm, show that in labyrinth, each cell is to the probability metrics of target area:

When robot initial direction is upper, add up the probability metrics figure in the labyrinth of " upward direction " according to above-mentioned calculating and statistical;

In like manner, when robot initial direction is right, the probability metrics figure in the labyrinth of " upward direction " is turned clockwise 90 °, keep coordinate constant, only numerical value variation;

Robot initial direction be under time, the probability metrics figure in the labyrinth of " upward direction " is turned clockwise 180 ° identical, keep coordinate constant, only numerical value variation;

When robot direction is left, the probability metrics figure in the labyrinth of " upward direction " is rotated counterclockwise 90 ° identical, keep coordinate constant, only numerical value variation;

3) application of probability metrics figure

Establishment step 2) described in have Direction Probability distance map after, robot is from starting point maze lattice search labyrinth, crossing is run in advancing, according to the current direction of robot, utilize probability metrics figure, judge that the probability metrics value of which direction maze lattice is minimum, decide course with this, until robot arrives at target area G.