CN104298239A - Enhanced map learning path planning method for indoor mobile robot - Google Patents

Abstract

The invention discloses an enhanced map learning path planning method for an indoor mobile robot. The method includes the steps that (1) ambient environment information is acquired, and an obstacle probability density model is established; (2) path planning is carried out through a greedy algorithm and an enhanced map learning method; (3) path selection and self-adaptation speed strategy adjustment are conducted on the indoor mobile robot. By the adoption of the enhanced map learning path planning method, a current optimal path can be planned in real time according to the current condition of the indoor mobile robot and the inherent non-holonomic constraint of the robot, meanwhile, the obstacle crossing ability, the target point convergence ability and the planning efficiency of the indoor mobile robot can be considered through the self-adaptation speed strategy adjustment, and therefore the indoor mobile robot can arrive at a specific location safely and effectively.

Description

A kind of indoor mobile robot strengthens map study paths planning method

Technical field

The present invention relates to the independent navigation field of ground wheeled robot, particularly a kind of indoor mobile robot strengthens map study paths planning method.

Background technology

Along with the development of Robotics and deepening continuously of artificial intelligence study, intelligent robot plays the part of more and more important role in human lives.As the one of common life robot, indoor mobile robot is used for indoor moving exhibition, home services as the substitute of attendant, the complex dynamic environments such as lounge guiding.In this kind of environment, environmental information destructuring, static dynamic barrier is staggered to be existed, and obviously, the ability to work of these factors to indoor mobile robot proposes challenge and requirement greatly in environmental information change.For completing service role preferably, indoor mobile robot needs to have detecting obstacles thing, and Division identification barrier, plans feasible path in real time, the ability of stability contorting action.Along with the development of sensor technology, computer technology and the network communications technology, real-time route plans that the brain as intelligent robot becomes the most important thing of indoor mobile robot research.

Indoor mobile robot task in the environment can be understood as from where coming? where go? how to go? what is done? and the environmental information that path planning can plan as a whole sensor acquisition makes the path decision meeting robot the present situation most, namely how to go.Conventional path planning can distinguish static programming and dynamic programming, static programming refers to that robot has possessed global context information, the path planning that robot is optimum under obtaining global context by calculated off-line, and active path planning spininess is to the routing in Dynamic Unknown Environment, robot understands limited and environment to environmental information and may change, and has the similarity of height with truth.For indoor mobile robot, we expect that it can realize real-time optimal path selection in indoor moving dynamic environment and dynamic barrier is hidden.

As controlling and the object of planning, indoor mobile robot is that a class becomes when having, the MIMO nonlinear systems of strong coupling and incomplete property.Due to the changeable complexity of environment and need consider more multifactor, its programmed decision-making become very complicated.In existing technology, conventional path planning, as fuzzy programming, genetic algorithm, ant group algorithm, neural network etc., often can not meet the requirement of dynamic environment and real-time simultaneously.In addition, the incomplete property that wheeled mobile robot exists also governs the routing of indoor mobile robot.Therefore, the main trend that the path planning algorithm with learning ability becomes present stage real-time active path planning research is studied.And design a kind of simple and reliable, real-time good, the indoor mobile robot planing method of being convenient to realize, can deal with multiclass Dynamic Unknown Environment is Deterministic service gordian technique of effectively carrying out working properly and realistic problem.

Summary of the invention

The present invention is directed in above-mentioned prior art the requirement being difficult in present paths planning method simultaneously meet Dynamic Unknown Environment planning and planning in real time, adopt the paths planning method strengthening map study, along with the movement of indoor mobile robot, constantly sharpen understanding the new ambient condition information obtained, the cost function in iterative computation random selecting path, study calculates the path of current time optimum, ensure that and well keep away barrier performance under the static-obstacle of indoor mobile robot, and the real-time planning function met when dynamic disorder occurs, reach the independent navigation and paths planning method that have compared with the indoor mobile robot of high-intelligentization.

A kind of indoor mobile robot strengthens map study paths planning method, comprises following step:

Step 1: set up the probability model that search coverage affects by barrier;

First, by the sonar sensor that indoor mobile robot is self-contained, obtain the ambient condition information of indoor mobile robot; Secondly, using indoor mobile robot the region of process as search coverage, the probability model that search coverage affects by barrier is set up according to described ambient condition information, and according to the ambient condition information real-time update probability model that affects by barrier of search coverage of sonar sensor Real-time Collection;

The probability model that described search coverage affects by barrier is as follows:

$F(X,Y)=1-\underset{i=1}{\overset{M}{\mathrm{\Π}}}[1-f_i(X,Y)]$

Wherein, the set of indoor moving indoor mobile robot work space information is described spatial information comprises all target locations and all Obstacle Positions; In current investigative range, work space information set is the work space information set detected is

Relative distance information between the barrier that indoor mobile robot is arrived by sonar radar detection and robot current location, the mileage gauge utilizing indoor mobile robot self-contained and inertial navigation system, relative distance information being converted to initial position is the positional information fastened of the base coordinate of initial point and range information.

{ (X, Y) } is search coverage, ? map on detected M barrier, fi (X, Y) is the influence function to indoor mobile robot routing of i-th barrier, adopt normal distribution represent as follows:

$f_i(X,Y)=\frac{1}{\sqrt{2\mathrm{\π}}{\mathrm{\σ}}_i}{e}^{(-\frac{{D_i}^2}{2{\mathrm{\σ}}_i})}$

Wherein, σ _ibe the coverage coefficient of i-th barrier, span is [0,1]; D _ibe i-th barrier to the distance matrix of all positions in search coverage, matrix size and map in the same size be N*N, each element in distance matrix is the physical distance of each position on barrier to map;

Step 2: based on greedy algorithm and enhancing study iterative strategy, at the maximum iteration time k of setting _maxin, iteration upgrades current location p _{now (t)}with target location p _goalbetween path cost function, using path corresponding to the path cost function reaching convergence as the optimal path of current time, detailed process is as follows:

Step 2.1: make k=1, k represents iterations;

Step 2.2: judge whether current iteration number of times has exceeded the maximum iteration time of setting, if exceed, then returns step 2.1; Otherwise, enter step 2.3;

Return step 2.1 to refer to from first time iteration, after rebuilding initial path, again enter iterated search optimal path;

Step 2.3: from the known map that the ambient condition information obtained is corresponding, Stochastic choice position pk inserts indoor mobile robot place current location p _{now (t)}with target location p _goal, wherein, p _ksatisfy condition pr} be before path point set under iteration optimal path cost function, for the set of indoor mobile robot work space information; Calculate the path cost function that kth time iteration obtains as follows $G_{p_{\mathrm{now}\left(t\right)}}^{k+1}=\min \{G_{p_{\mathrm{now}\left(t\right)}}^k,R_{p_{\mathrm{now}\left(t\right),p_k}}+G_p^k\};$

Above-mentioned formula is utilize the path cost function computing formula strengthening study mechanism and obtain;

Wherein, the path cost function that kth-1 iteration obtains, and for current location p on known map _{now (t)}with selected location p _kbetween connection weights, obtain for adopting greedy algorithm obtain after kth time iteration completes from position p _k-1to target location p _goaloptimal path cost function:

$G_p^k=\left\{\begin{array}{cc}{R}_{p_k,p_{\mathrm{goal}}}+{G_p}_{\mathrm{goal}}& k=1\\ \min \{R_{p_k,p_{\mathrm{goal}}}+{G_p}_{\mathrm{goal}},R_{p_k,p_{k-1}+G_p^{k-1}\}}& k\≠1\end{array}\right.$

Wherein, for target location p _goalcost function, and for on known map at kth time selected location point p _kwith kth-1 selected location p _k-1between connection weights;

Any two positions p in known map _aand p _bbetween connection weights affect probability by the barrier that the air line distance between two positions and this straight line are passed to form: $R_{p_a,p_b}=d_{p_a,p_b}+\max \left(\underset{C_{p_a,p_b}}{\∫}F(X,Y)\mathrm{ds}\right)$

Wherein, for position p _ato position p _bphysical location distance between 2; for position p _ato position p _bpath point set, ds is path integral unit; Max () finds a function for maximal value, namely tries to achieve position p _ato position p _bset of paths in the maximum probability that affects by barrier of a certain path;

Step 2.4: the path cost function obtained after judging iteration whether reach convergence, if convergence, then exit iterative process, path corresponding for the path cost function of having restrained is entered step 3 as optimal path; Otherwise, iterations k is added 1, returns step 2.2;

Step 3: the optimal path obtained using step 2, as the current preselected path of indoor mobile robot, judges the deflection angle φ of indoor mobile robot according to the position of preselected path and indoor mobile robot and velocity _twhether meeting wheeled robot can not the nonholonomic constraint of lateral slip, determines whether indoor mobile robot moves according to preselected path direction:

If do not meet, then return step 2, rebuild all paths of current time; Due in step 2.3, insert indoor mobile robot place current location p _{now (t)}with target location p _goalbetween position p _kfor Stochastic choice, wherein, p _ksatisfy condition { p _rbe path point set before under iteration optimal path cost function, for the set of indoor mobile robot work space information.Therefore, after returning step 2, the insertion position that Stochastic choice obtains, can be different from the insertion position selected before, can obtain new path;

If meet, move with the optimal path direction that the indoor mobile robot translational speed of setting is cooked up along current time, enter the path planning of subsequent time, t=t+1, return step 1, until indoor mobile robot moves to target location, complete path planning;

The deflection angle φ of described indoor mobile robot _twhether meet wheeled robot can not the nonholonomic constraint of lateral slip refer to: φ _t∈ [0,60 °] ∪ [120 °, 180 °];

Wherein, p _tand p _t-1be respectively indoor mobile robot current location and a upper moment position, for in t, indoor mobile robot is by planning first path point in the optimal path that obtains, and namely indoor mobile robot is at the pre-arriving at location in t+1 moment.

P _t-1by p _t-1=p _t-vt obtains, and v is the indoor mobile robot translational speed of setting.

Indoor mobile robot place current location p is inserted in described step 2 _{now (t)}with target location p _goalbetween position p _kbasis for selecting as follows:

1) when target location appears in the current investigative range of indoor mobile robot then using target location as the position chosen during the 1st iteration: p ₁=p _goal; From the 2nd iteration and k=2, p _kfor the position of random selecting in body of a map or chart, and need satisfy condition wherein, { p _rbe path point set before under iteration optimal path cost function, for the set of indoor mobile robot work space information;

2) when target location does not appear in the current investigative range of indoor mobile robot, then position p in each iterative process _kfor the position of random selecting from known map.

In described step 1, sonar sensor entrained by indoor mobile robot self refers to the sonar contact radar that adopted chassis Pioneer-2DX is self-contained.

Any two positions p in described known map _aand p _bbetween connection weights calculate as follows:

$R_{p_a,p_b}=d_{p_a,p_b}+k_m\×\max \left(\underset{C_{p_a,p_b}}{\∫}F(X,Y)\mathrm{ds}\right),$

Wherein, k _mfor barrier affect amplification coefficient be conducive to improve robot keep away barrier effect, span is [50,1000].

In described step 4, the indoor mobile robot v of setting carries out self-adaptative adjustment as follows:

Wherein, p _barrierfor Obstacle Position, v ₀for the basal rate setting value of indoor mobile robot movement, for work space information set in current investigative range, comprise the Obstacle Position in current investigative range, aiming spot, and moving area map position information.

Beneficial effect

Compared with prior art, the invention has the advantages that:

1, the present invention adopts the probabilistic Modeling of map barrier, the barrier detected by indoor mobile robot is built by the mode of probability affects model, make barrier form continuous print impact to surrounding environment, and this impact is only for the region that indoor mobile robot is probed;

2, the present invention is compared with in the past traditional mobile robot path planning, by adding map learning ability to indoor service, robot is enable to carry out analytic learning to ever-increasing cartographic information, path planning best under obtaining current location by greedy algorithm study.

3, invention provides for nonholonomic constraint when indoor mobile robot is planned and adaptive speed shift strategy, not only solve the real system application problem of path planning algorithm, and taken into account planning efficiency and the mobile security of indoor mobile robot.

Accompanying drawing explanation

Fig. 1 is the process flow diagram of the method for the invention;

Fig. 2 is indoor mobile robot hardware system structure schematic diagram;

Fig. 3 is indoor mobile robot nonholonomic constraint schematic diagram;

Fig. 4 adopts the path planning design sketch of the method for the invention under dynamic and static state environment, and wherein, figure (a) strengthens design sketch when map study path planning hides dynamic disorder; Figure (b) strengthens design sketch when map study path planning hides static-obstacle; Figure (c) strengthens design sketch when map study path planning finally completes.

Embodiment

Below with reference to accompanying drawing and concrete case study on implementation, the present invention will be further described.

As shown in Figure 1, the present invention is that a kind of indoor mobile robot strengthens map study paths planning method, comprises following step:

Step 1: set up the probability model that search coverage affects by barrier;

First, by the sonar sensor that indoor mobile robot is self-contained, obtain the ambient condition information of indoor mobile robot; Secondly, using indoor mobile robot the region of process as search coverage, the probability model that search coverage affects by barrier is set up according to described ambient condition information, and according to the ambient condition information real-time update probability model that affects by barrier of search coverage of sonar sensor Real-time Collection;

In described step 1, sonar sensor entrained by indoor mobile robot self refers to the sonar contact radar in eight directions, front that adopted chassis Pioneer-2DX is self-contained, as shown in Figure 2.According to the obstacle information that sonar sensor detects, the barrier impact probability model building probed region is as follows:

$F(X,Y)=1-\underset{i=1}{\overset{M}{\mathrm{\Π}}}[1-f_i(X,Y)]---\left(1\right)$

Wherein, the set of indoor mobile robot work space information is described spatial information comprises all target locations and all Obstacle Positions; In current investigative range, work space information set is the work space information set detected is

[the relative distance information between the barrier that indoor mobile robot is arrived by sonar radar detection and robot current location, the mileage gauge utilizing indoor mobile robot self-contained and inertial navigation system, relative distance information being converted to initial position is the positional information fastened of the base coordinate of initial point and range information.】

{ (X, Y) } is search coverage, ? map on detected M barrier, f _i(X, Y) is the influence function to indoor mobile robot routing of i-th barrier, adopts normal distribution to represent as follows:

$f_i(X,Y)=\frac{1}{\sqrt{2\mathrm{\π}}{\mathrm{\σ}}_i}{e}^{(-\frac{{D_i}^2}{2{\mathrm{\σ}}_i})}---\left(2\right)$

Wherein, σ _ibe the coverage coefficient of i-th barrier, span is [0,1]; D _ibe i-th barrier to the distance matrix of all positions in search coverage, matrix size and map in the same size be N*N;

Step 2: based on greedy algorithm and enhancing study iterative strategy, at the maximum iteration time k of setting _maxin, iteration upgrades current location p _{now (t)}with target location p _goalbetween path cost function, using path corresponding to the path cost function reaching convergence as the optimal path of current time, detailed process is as follows:

Step 2.1: make k=1, k represents iterations;

Step 2.2: judge whether current iteration number of times has exceeded the maximum iteration time of setting, if exceed, then returns step 2.1; Otherwise, enter step 2.3;

Return step 2.1 to refer to from first time iteration, after rebuilding initial path, again enter iterated search optimal path;

Step 2.3: from the known map that the ambient condition information obtained is corresponding, Stochastic choice position p _kinsert indoor mobile robot place current location p _{now (t)}with target location p _goal, calculate the path cost function that kth time iteration obtains as follows

$G_{p_{\mathrm{now}\left(t\right)}}^{k+1}=\min \{G_{p_{\mathrm{now}\left(t\right)}}^k,R_{p_{\mathrm{now}\left(t\right),p_k}}+G_p^k\}---\left(3\right)$

Wherein, the path cost function that kth-1 iteration obtains, and for current location p on known map _{now (t)}with selected location p _kbetween connection weights, obtain for adopting greedy algorithm obtain after kth time iteration completes from position p _k-1to target location p _goaloptimal path cost function:

$G_p^k=\left\{\begin{array}{cc}{R}_{p_k,p_{\mathrm{goal}}}+{G_p}_{\mathrm{goal}}& k=1\\ \min \{R_{p_k,p_{\mathrm{goal}}}+{G_p}_{\mathrm{goal}},R_{p_k,p_{k-1}+G_p^{k-1}\}}& k\≠1\end{array}\right.---\left(4\right)$

Wherein, for target location p _goalcost function, and for on known map at kth time selected location point p _kwith kth-1 selected location p _k-1between connection weights;

Any two positions p in known map _aand p _bbetween connection weights affect probability by the barrier that the air line distance between two positions and this straight line are passed to form:

$R_{p_a,p_b}=d_{p_a,p_b}+k_m\×\max \left(\underset{C_{p_a,p_b}}{\∫}F(X,Y)\mathrm{ds}\right)---\left(5\right)$

Wherein, for position p _ato position p _bphysical location distance between 2; for position p _ato position p _bpath point set, ds is path integral unit; Max () finds a function for maximal value;

Step 2.4: the path cost function obtained after judging iteration whether reach convergence, if convergence, then exit iterative process, path corresponding for the path cost function of having restrained is entered step 3 as optimal path; Otherwise, iterations k is added 1, returns step 2.2;

Step 3: the optimal path obtained using step 2, as the current preselected path of indoor mobile robot, judges the deflection angle φ of indoor mobile robot according to the position of preselected path and indoor mobile robot and velocity _twhether meeting wheeled robot can not the nonholonomic constraint of lateral slip, determines whether indoor mobile robot moves according to preselected path direction:

If do not meet, then return step 2, rebuild all paths of current time; Due in step 2.3, insert indoor mobile robot place current location p _{now (t)}with target location p _goalbetween position p _kfor Stochastic choice, wherein, p _ksatisfy condition { p _rbe path point set before under iteration optimal path cost function, for the set of indoor mobile robot work space information.Therefore, after returning step 2, the insertion position that Stochastic choice obtains, can be different from the insertion position selected before, can obtain new path;

If meet, move with the optimal path direction that adaptive speed is cooked up along current time, enter the path planning of subsequent time, t=t+1, return step 1, until indoor mobile robot moves to target location, complete path planning;

Generally speaking, because wheeled mobile robot cannot be displaced sideways, can suppose that wheeled mobile robot nothing in motion process is skidded, only make pure rolling.Namely think that robot does not have component motion on its transverse axis, this nonholonomic constraint can be expressed as:

$\stackrel{\·}{x}\sin \mathrm{\θ}-\stackrel{\·}{y}\cos \mathrm{\θ}=0$

The deflection angle φ of described indoor mobile robot _twhether meet wheeled robot can not the nonholonomic constraint of lateral slip refer to: φ _t∈ [0,60 °] ∪ [120 °, 180 °], as shown in Figure 3;

Wherein, p _tand p _t-1be respectively indoor mobile robot current location and a upper moment position, for in t, indoor mobile robot is by planning first path point in the optimal path that obtains, and namely indoor mobile robot is at the pre-arriving at location in t+1 moment.

Indoor mobile robot moves with adaptive speed v:

Algorithm performance is analyzed

1, convergence

In study iterative strategy, by formula (3), (4) are known if iterations is enough large, to finally converge to a stable point.

Lemma 1 is for any point p in map _i, i ∈ N ²with impact point p _goalbetween the cost function in path be one to increase and the sequence of monotone decreasing with iterations k.

From formula (4), the map location p chosen for the first time in step 2 ₁using when first time iteration as the insertion position between current location and target location, form initial path.From second time iteration, new random selecting position after insertion, path cost function by below two select in choose minimum one:

(1) starting point---new insertion position p _k---the path line of impact point.

(2) starting point---new insertion position p _k---the best cost function path { p of iteration before _r| the path line of r ∈ k}.

Theorem 1 supposes that the connection weights R perseverance of any two positions is just, and the meeting of iteration produces the initial path cost function of current location and target location for the first time utilize formula (3) and formula (4) to upgrade path cost function and there is several character below.

1, the path cost sequence of function of optional position and target location can final convergence.

2, the cost function of target location for optional position on map and path cost function between target location to keep stable after certain iterations.

3, from impact point p _goalstart, one finds starting point surely in limited step.

Prove:

Character 1: from formula (5), the distance between any two diverse locations i, j ∈ N ²for just, the accumulated probability on path is affecting ∫ _cf (X, Y) ds is also positive number, and therefore, the connection weights R of any two diverse locations is positive number, and cost function G is also positive number.According to lemma 1, for any one p _{now (t)}and the cost function sequence in path between target location along with the increase of iterations will converge on minimum border.

Character 2: by the cost function of the known impact point of cost function initialization perseverance is 0, and do as one likes matter 1 is known, to finally restrain, if iterations is greater than the number of position on map, then the full map of traversal is chosen in the path between current location and the position of impact point, namely gets optimum solution, and then cost function is minimally constant to continue iteration.

Character 3: according to learning strategy, choose current point and path between impact point after, what comprise in this path chooses random site sequence { p _r| r ∈ k} is inadvisable in next iteration.Therefore, after certain iterations, all positions in map all can be selected, and comprise starting point.

2, complicacy analysis

In order to represent the complexity of algorithm, defining variable T represents the time (step number) of path finding, and concrete analysis of complexity is as follows:

(1) if definition two-dimensional map size is N × N, then the connection weights between all positions of full map are calculated complexity be N ⁴, wherein, i, j ∈ N ², i ≠ j.

(2) in learning strategy, robot passes through the path cost function that constantly searching each moment is minimum the complexity of greedy search algorithm used herein is wherein for the averaging time of path finding.

(3) path completing t when robot is chosen, and after entering the t+1 moment, the connection weights complexity upgrading map is N ⁴.

Therefore, indoor mobile robot can be expressed as from the Path complexity of origin-to-destination:

Below with a concrete application example, operation of the present invention is described in detail, enhancing map study path planning algorithm of the present invention is mainly used in the independent navigation of indoor mobile robot and keeps away in barrier planning, arrives at here embody its performance mainly for the impact point in Static and dynamic two class obstacle environment.Specifically arrange as follows:

Experimental situation is set to grating map, map size N=50m, has dynamic and static state two class obstacle in map concurrently, dynamic disorder is initially located at (5,10), with speed 1m/s in (5,10) and (15,10) back and forth movement between, static-obstacle thing is L shape obstacle, and wherein three summits of static-obstacle thing 1 are respectively (18,20), (25,20), (25,10), three summits of static-obstacle thing 2 are respectively (30,40), (30,35), (38,35).Eight sonar sensors that indoor mobile robot can only be carried by its chassis obtain the map obstacle information in eight directions, mobile front.Barrier affects amplification coefficient k _m=500, barrier coverage σ _i=1, i=1,2 .., M, the investigative range R of indoor mobile robot _detect=10m, basic movement speed v ₀=1m/s, indoor mobile robot starting point is (0,0), and impact point is (48,48).

As shown in Figure 4, give indoor moving indoor mobile robot and have the real-time active path planning track in dynamic and static state environment concurrently.Wherein, scheme (a) and strengthen effect when map study path planning hides dynamic disorder; Figure (b) strengthens effect when map study path planning hides static-obstacle; Figure (c) strengthens effect when map study path planning finally completes; " * " represents robot mobile route, and "○" represents barrier, and the encapsulated coil around barrier represents barrier probability density, and the closer to barrier, probability density is larger, and namely the closing line number of turns is more, and " " represents target location.As can be seen from the figure, after providing target location, although indoor moving indoor mobile robot obtains incomplete to the obstacle information in environment, and there is dynamic disorder, it relies on enhancing map learning algorithm can realize good target and arrives at and barrier avoiding function, has good planning efficiency and barrier avoiding function.

Claims (5)

1. indoor mobile robot strengthens a map study paths planning method, it is characterized in that, comprises following step:

Step 1: set up the probability model that search coverage affects by barrier;

First, by the sonar sensor that indoor mobile robot is self-contained, obtain the ambient condition information of indoor mobile robot; Secondly, using indoor mobile robot the region of process as search coverage, the probability model that search coverage affects by barrier is set up according to described ambient condition information, and according to the ambient condition information real-time update probability model that affects by barrier of search coverage of sonar sensor Real-time Collection;

The probability model that described search coverage affects by barrier is as follows:

$F(X,Y)=1-\underset{i=1}{\overset{M}{\mathrm{\Π}}}[1-f_i(X,Y)]$

Wherein, the set of indoor mobile robot work space information is described spatial information comprises all target locations and all Obstacle Positions; In current investigative range, work space information set is the work space information set detected is

{ (X, Y) } is search coverage, ? map on detected M barrier, f _i(X, Y) is the influence function to indoor mobile robot routing of i-th barrier, adopts normal distribution to represent as follows:

$f_i(X,Y)=\frac{1}{\sqrt{2\mathrm{\π}}{\mathrm{\σ}}_i}{e}^{(-\frac{{D_i}^2}{{2\mathrm{\σ}}_i})}$

Wherein, σ _ibe the coverage coefficient of i-th barrier, span is [0,1]; D _ibe i-th barrier to the distance matrix of all positions in search coverage, matrix size and map in the same size be N*N;

Step 2: based on greedy algorithm and enhancing study iterative strategy, at the maximum iteration time k of setting _maxin, iteration upgrades current location p _{now (t)}with target location p _goalbetween path cost function, using path corresponding to the path cost function reaching convergence as the optimal path of current time, detailed process is as follows:

Step 2.1: make k=1, k represents iterations;

Step 2.2: judge whether current iteration number of times has exceeded the maximum iteration time of setting, if exceed, then returns step 2.1; Otherwise, enter step 2.3;

Step 2.3: from the known map that the ambient condition information obtained is corresponding, Stochastic choice position p _kinsert indoor mobile robot place current location p _{now (t)}with target location p _goal, wherein, p _ksatisfy condition { p _rbe path point set before under iteration optimal path cost function, for the set of indoor mobile robot work space information; Calculate the path cost function that kth time iteration obtains as follows

Wherein, the path cost function that kth-1 iteration obtains, and for current location p on known map _{now (t)}with selected location p _kbetween connection weights, obtain for adopting greedy algorithm obtain after kth time iteration completes from position p _kto target location p _goaloptimal path cost function:

$G_p^k=\left\{\begin{array}{cc}{R}_{p_k,p_{\mathrm{goal}}}+G_{p_{\mathrm{goal}}}& k=1\\ \min \{R_{p_k,p_{\mathrm{goal}}}+G_{p_{\mathrm{goal}}},R_{p_k,p_{k-1}}+G_p^{k-1}\}& k\≠1\end{array}\right.$

Wherein, for target location p _goalcost function, and for on known map at kth time selected location point p _kwith kth-1 selected location p _k-1between connection weights;

Any two positions p in known map _aand p _bbetween connection weights affect probability by the barrier that the air line distance between two positions and this straight line are passed to form: $R_{p_a,p_b}=d_{p_a,p_b}+\max \left(\underset{C_{p_a,p_b}}{\∫}F(X,Y)\mathrm{ds}\right)$

Wherein, for position p _ato position p _bphysical location distance between 2; for position p _ato position p _bpath point set, ds is path integral unit; Max () finds a function for maximal value;

Step 2.4: the path cost function obtained after judging iteration whether reach convergence, if convergence, then exit iterative process, path corresponding for the path cost function of having restrained is entered step 3 as optimal path; Otherwise, iterations k is added 1, returns step 2.2;

Step 3: the optimal path obtained using step 2, as the current preselected path of indoor mobile robot, judges the deflection angle φ of indoor mobile robot according to the position of preselected path and indoor mobile robot and velocity _twhether meeting wheeled robot can not the nonholonomic constraint of lateral slip, determines whether indoor mobile robot moves according to preselected path direction:

If do not meet, then return step 2; If meet, move with the optimal path direction that the indoor mobile robot translational speed of setting is cooked up along current time, enter the path planning of subsequent time, t=t+1, return step 1, until indoor mobile robot moves to target location, complete path planning;

The deflection angle φ of described indoor mobile robot _twhether meet wheeled robot can not the nonholonomic constraint of lateral slip refer to: φ _t∈ [0,60 °] ∪ [120 °, 180 °];

Wherein, p _tand p _t-1be respectively indoor mobile robot current location and a upper moment position, for in t, indoor mobile robot is by planning first path point in the optimal path that obtains, and namely indoor mobile robot is at the pre-arriving at location in t+1 moment;

P _t-1by p _t-1=p _t-vt obtains, and v is the indoor mobile robot translational speed of setting.

2. indoor mobile robot according to claim 1 strengthens map study paths planning method, it is characterized in that, inserts indoor mobile robot place current location p in described step 2 _{now (t)}with target location p _goalbetween position p _kbasis for selecting as follows:

1) when target location appears in the current investigative range of indoor mobile robot then using target location as the position chosen during the 1st iteration: p ₁=p _goal; From the 2nd iteration and k=2, p _kfor the position of random selecting in body of a map or chart, and need satisfy condition wherein, { p _rbe path point set before under iteration optimal path cost function, for the set of indoor mobile robot work space information;

2) when target location does not appear in the current investigative range of indoor mobile robot, then position p in each iterative process _kfor the position of random selecting from known map.

3. indoor mobile robot according to claim 2 strengthens map study paths planning method, it is characterized in that, in described step 1, sonar sensor entrained by indoor mobile robot self refers to the sonar contact radar that adopted chassis Pioneer-2DX is self-contained.

4. indoor mobile robot according to claim 3 strengthens map study paths planning method, it is characterized in that, any two positions p in described known map _aand p _bbetween connection weights calculate as follows:

$R_{p_a,p_b}=d_{p_a,p_b}+k_m\×\max \left(\underset{C_{p_a,p_b}}{\∫}F(X,Y)\mathrm{ds}\right);$

Wherein, k _mfor barrier affect amplification coefficient be conducive to improve robot keep away barrier effect, span is [50,1000].

5. the indoor mobile robot according to any one of claim 1-4 strengthens map study paths planning method, and it is characterized in that, in described step 4, the indoor mobile robot v of setting carries out self-adaptative adjustment as follows:

Wherein, p _barrierfor Obstacle Position, v ₀for the basal rate setting value of indoor mobile robot movement, for work space information set in current investigative range, comprise the Obstacle Position in current investigative range, aiming spot, and moving area map position information.