CN101077578A - Mobile Robot local paths planning method on the basis of binary environmental information - Google Patents

Abstract

The process of programming local path of mobile robot based on binary environmental information includes the following steps: 1. calculating the coordinates of the target point in the local path programming coordinate system; 2. calculating the default path without barrier; 3. initializing the intermediate variables; 4. obtaining binary environmental information description based on the output of the sensor and classifying the barriers; 5. selecting current optimal arc section; 6. calculating the relative position relation between carriers and the current optimal arc section, and re-classifying the barriers; 7. renewing the right and left barrier bypassing arc sections, and backing to the step 4 for the next loop; and 8. outputting the local path programming results. The present invention is simple and high in efficiency, and can meet the real-time programming requirement for continuous walking of mobile robot.

Description

A kind of mobile robot's local paths planning method based on the binary environmental information

Technical field

The present invention relates to a kind of local paths planning method, be applicable to a class running on wheels mobile robot local paths planning.

Background technology

The task of mobile robot path planning is to seek one from the suitable motion path of given origin-to-destination in the working environment of barrier is arranged, and robot can not had safely in motion process walk around all barriers with bumping.Mobile robot path planning mainly solves three problems: (1) makes robot move to impact point from initial point; (2) make robot can get around barrier and some point that must pass through of process with certain algorithm; (3) optimize the robot running orbit under the prerequisite of above task finishing as far as possible.

Can be divided into two kinds according to working environment path planning model: based on the global path planning of model, the full detail of operating environment is known, claims static state or off-line path planning again; Sensor-based local paths planning, operating environment information are all unknown or part is unknown, claim dynamic or online path planning again.

Local paths planning only need be nearer apart from the mobile robot obstacle information.Robot is in moving process, constantly upgrade its inner environment representation according to the information of sensor, thereby determine the barrier distribution situation in the subrange around the current location in map reaches, and on this basis, with final arrival impact point is purpose, cooks up a local optimum path.

The travelling performance of the design of local paths planning method or selection and robot, environment essential characteristic, robot sensor configuration, the requirement of sector planning real-time etc. are all closely related, also do not have a general local paths planning method that is applicable to any condition at present.

Applicable object of the present invention is a class wheeled mobile robot, also can be described as intelligent vehicle, its characteristics are: (1) mobile device adopts wheeled construction, possess along straight line or arc direction and advance or retreat and the pivot turn ability, but in moving process, for keeping the continuous and continuity of motion, requiring its mobile route usually is continuous curve, and the curvature of any point is avoided walking with broken line form greater than certain definite value on the curve; (2) this mobile robot's main running region is based on more smooth open ground, and is distributed with the obstacle that varies in size, that is, movement environment is made up of barrier and free space, and the quantity of barrier and size are all limited; (3) complaint message of the sensor that robot disposed in can the perception certain limit, comprise obstacle size and with the relative position of robot; (4) the local paths planning real-time of robot is had relatively high expectations, and should possess the ability of keeping away major obstacle in continuous traveling process.

Using more local paths planning method at present comprises: Artificial Potential Field method, fuzzy logic algorithm, neutral net method, genetic algorithm and grid method etc.All there is different shortcomings in these methods in application, the Artificial Potential Field method is only considered the influence to path planning of impact point and obstacle, and the path of planning output shows as broken line form more, and the locomotivity of robot is had higher requirement; Fuzzy logic algorithm has been considered the experience of car steering, is more suitable for being applied to existing the complex environment of condition such as moving disorder, seems comparatively complicated for general binary environment; Overall planing methods such as grid method also can be applied to local paths planning after transforming, but sensor is had relatively high expectations, and amount of calculation is bigger, and real-time is not strong.

In most algorithm researches, all pay close attention to the generative process of path planning more, and to the judgement of complaint message in the planning process with handle not quite clear, and in actual applications, when the design of algorithm and programming, how to be rich in orderliness ground more efficiently and to handle complaint message, also be improve executing efficiency, make code more succinctly understandable, be convenient to safeguard the problem that institute must consideration.

Therefore, be necessary to design a kind of simple efficient and be easy to the local paths planning algorithm of programming and realizing, under the basic assumption prerequisite of satisfied rather level areas shape on a large scale, can plan that the mobile robot gets around obstacle, (promptly avoid broken line and retreat, only advance) the arrival impact point of advancing continuously, smoothly along continuous smooth curve.

Summary of the invention

Technology of the present invention is dealt with problems: overcome the deficiencies in the prior art, provide a kind of simple, efficient, the mobile robot's local paths planning method based on the binary environmental information that is easy to realize has satisfied the planning real-time requirement of mobile robot's continuous walking.

Technical solution of the present invention: a kind of mobile robot's local paths planning method based on the binary environmental information is characterized in that step is as follows:

(1) calculates the coordinate (x of impact point under the local paths planning coordinate system ₀, y ₀);

Acquiescence path planning when (2) calculating is accessible comprises that acquiescence output curvature is S ₀With acquiescence output arc length L ₀

(3) intermediate variable, promptly barrier segmental arc and local paths planning parameter initialization are kept away in the left and right sides

The barrier segmental arc is kept away for kept away the required minimum curvature of obstacle on the current impact point direction from the left side in a definition left side, i.e. segmental arc the most on the right side, the barrier segmental arc is kept away for kept away the required maximum curvature of current obstacle from the right side in the right side, i.e. the segmental arc in close left side, about keep away the barrier segmental arc arc length be respectively L _l, L _r, about keep away the barrier segmental arc curvature be respectively S _l, S _r, under the primary condition, two segmental arcs all overlap with the acquiescence planning segmental arc that calculates in the step (2), and S is promptly arranged _l=S _r=S ₀, L _l=L _r=L ₀

The parameter of local paths planning is: mobile robot's length a, width b, the min. turning radius R that the mobile robot can realize _MinOr maximum curvature S _Max, local paths planning coverage D, R _MinAnd S _MaxBe scalar, and satisfy S _Max=1/R _Min, the output segmental arc curvature of local paths planning is limited in [S _Max, S _Max] in the interval, D=min (D _s, D _t), D wherein _sBe the environment sensing coverage, can determine D according to sensor configuring condition and motion control precision _tBe the impact point distance

(4), obtain the binary environmental information and describe, and the complaint message in the binary environmental information is carried out preliminary classification according to sensor output

A. with the complaint message in the binary environment with x _Fi, x _Ri, y _Li, y _RiForm provide, the respectively preceding border X coordinate of corresponding each obstacle, border, back X coordinate, left margin Y coordinate, right margin Y coordinate, i=1,2 ..., n, n are the obstacle sum;

B. according to obstacle with about keep away the relative position relation of barrier segmental arc, obstacle is divided into four classes, promptly unknown obstacle, left side obstacle, right side obstacle, can ignore obstacle; Unknown obstacle be obstacle with about keep away relative position relation the unknown of barrier segmental arc; The left side obstacle is that obstacle is positioned at the right side and keeps away barrier segmental arc left side; The right side obstacle is that obstacle is positioned at a left side and keeps away barrier segmental arc right side; Can ignore obstacle and comprise two class situations, one be obstacle about keep away between the barrier segmental arc, then can think that this obstacle was kept away this moment, can not consider, its two for obstacle to mobile robot's distance greater than coverage, can not influence the execution of this route programming result, therefore in current local paths planning process, can not consider;

C. all obstacle preliminary classification are defined as unknown obstacle;

(5) choose current optimum segmental arc

Keep away the barrier segmental arc relatively, immediate from wherein choosing curvature with the curvature of acquiescence path planning, be considered as current optimum segmental arc, the curvature and the arc length of optimum segmental arc are respectively S _cAnd L _c, promptly compare | S _l-S ₀| and | S ₀-S _r|, if | S _l-S ₀|≤| S ₀-S _r|, then make S _c=S _l, L _c=L _l, otherwise make S _c=S _r, L _c=L _r

(6) relative position relation of the current optimum segmental arc of dyscalculia and step (5), and obstacle reclassified

Order reads the information of obstacle, if unknown obstacle and obstacle and mobile robot's distance then directly can be ignored obstacle with its heavy being categorized as greater than planning coverage D; Otherwise the relative position relation of dyscalculia and current optimum segmental arc, disturbance in judgement is on the left side or the right side of optimum segmental arc, or drop on the optimum segmental arc, if obstacle drops on the optimum segmental arc, then interrupt step (6) enters step (7), otherwise obstacle is reclassified, read next complaint message then; If all complaint messages have all read and have finished, still, be feasible planning segmental arc not because obstacle on segmental arc and interrupt, does not have obstacle on the then current optimum segmental arc, can directly change step (8) output program results over to;

(7) keep away the barrier segmental arc about the renewal

Current optimum segmental arc is that a left side is kept away when hindering segmental arc, then upgrades a left side and keeps away the barrier segmental arc; Current optimum segmental arc is that the right side is kept away when hindering segmental arc, then upgrade the right side and keep away the barrier segmental arc, about upgrading, keep away after the barrier segmental arc, obstacle no longer drops on the segmental arc, keeps away the barrier segmental arc for a left side, and this moment, obstacle was on the segmental arc right side, keep away the barrier segmental arc for the right side, this moment, obstacle was in segmental arc left side, and then according to the category attribute of obstacle, current obstacle reclassified;

After finishing above-mentioned work, return step (5) and choose optimum segmental arc again;

(8) output local paths planning result

If behind all obstacles of traversal, do not have obstacle to drop on the current optimum segmental arc, then current optimum segmental arc is feasible planning segmental arc, with curvature S in the step (6) _cWith arc length L _cExport to navigation and control system as current optimum programming path, and finish current local paths planning process.

The present invention's advantage compared with prior art is: than each known local paths planning method, the present invention is primarily aimed at the basic assumption of level terrain on a large scale, ignore of the influence of complicated rolling topography to the mobile robot, with the least possible data volume describe environment information, therefore simpler, the required planning time is shorter, can satisfy the planning real-time requirement of mobile robot's continuous walking; The reduction of environmental information data volume demand makes that the mobile robot need not to be equipped with the environment sensing demand that complicated multi-sensor can satisfy local paths planning, greatly reduces the hardware configuration requirement; For snagged environment, to the obstacle processing of simply classifying, improved executing efficiency, be easy to Project Realization especially.

Description of drawings

Fig. 1 local paths planning execution cycle of the present invention;

Fig. 2 is a local paths planning method flow chart of the present invention;

Fig. 3 is a local paths planning coordinate system schematic diagram of the present invention;

This acquiescence path planning when accessible of the present invention of Fig. 4 calculates;

Fig. 5 keeps away barrier segmental arc schematic diagram about among the present invention;

Fig. 6 is the relative position relation of obstacle of the present invention and segmental arc;

Fig. 7 keeps away the calculating of barrier segmental arc for a left side of the present invention;

Fig. 8 finishes and keeps away the barrier continuous walking and arrive impact point by repeatedly calling the local paths planning algorithm for the present invention.

The specific embodiment

In the autonomous walking process of mobile robot, need carry out local paths planning method at regular intervals one time, i.e. the present invention generates current local optimum path, offers navigation and control system.

The execution cycle of local paths planning method process of the present invention can be divided into dual mode: fixed cycle is carried out and variable period is carried out.Local paths planning process frequency was higher when fixed cycle was carried out, and was primary goal as far as possible accurately to obtain environmental information, to reduce erroneous judgement; Local paths planning process frequency dynamic changed when variable period was carried out, and was primary goal successfully to keep away known obstacle.Can be according to the different choice of application different forms.If environmental information of sensor perception not accurate enough (as the computational accuracy that has certain probability of miscarriage of justice, obstacle location not good enough etc.) or obstacle number are more, need carry out environmental data collecting continually, the frequency of local paths planning is higher, then should adopt the fixed cycle; If sensor sensing range limited (spending less than 180) as the viewing field of camera angle, and there is not comprehensive historical data (each local paths planning is a main task to keep away current obstacle that can perception only all) during computing environment information, then adopt the fixed cycle may form the perception blind area of obstacle, cause and keep away results such as barrier action is imperfect, make and keep away the barrier failure, therefore should select variable period for use.

The selection of fixed cycle or variable period can embody by the arc length of local optimum path segmental arc of output, i.e. navigation can be determined calling the frequency and the concrete moment of local paths planning process according to the arc length of output with control system.If mobile robot's travel speed is V, the total time of sensor data collection and processing is t, the time that the time of local paths planning handles with respect to sensor data can ignore, the output arc length is L, for realizing mobile robot's continuous walking, the maximum distance travelled of then calling the local paths planning process for twice should be chosen as L-Vt at interval, promptly, called the local paths planning process constantly from last time, mobile robot's distance travelled reaches before the L-Vt, promptly should start local paths planning next time.When L is definite value, local paths planning is that the fixed cycle execution (should be a set a distance strictly speaking, the startup that is the local paths planning process is not to determine according to running time, but determine according to the distance travelled number, in the actual motion process, mobile robot's speed is not strict constant, so the cycle of local paths planning neither strict conformance); When L was variate, local paths planning was that variable period is carried out.

As shown in Figure 1, start local paths planning the n time at point 1 place among the figure, receive the output of local paths planning to point 2 places, from putting 2 to putting till 4, the mobile robot advances (in like manner according to the curvature S and the arc length L of the n time local paths planning output, point 1 is to point 2 highway sections, and the mobile robot advances according to the output of the n-1 time local paths planning), and start local paths planning the n+1 time putting 3 places.1,2 and 3,4 distance is Vt, and 1,4 distance is L _n(the output L of the n time local paths planning), 3,6 distance is L _N+1(the n+1 time planning output L), 2,3 segment distances are L-2Vt.As seen from the figure, the arc length of planning output should be able to make that local paths planning received current route programming result in the past starting next time.Therefore, the minimum of local paths planning output arc length is 2Vt, and (L-2Vt)/L is big more, promptly to account for the proportion of 1,4 distances big more for 2,3 distance, the mobile robot can carry out the output of local paths planning more fully, reduces the error between path planning and the actual motion.When impact point and mobile robot's distance during, then need not to carry out local paths planning less than 2Vt.

The main implementation procedure of local paths planning method of the present invention is as shown in Figure 2:

(1) calculates the coordinate of impact point under the local paths planning coordinate system.

In mobile robot's traveling process, may there be one or more impact points, robot needs the overall situation planning according to the operator, sequentially arrives or all or part of impact point of approaching arrival.But for local paths planning, a certain moment can only consider on the moving direction that apart from a nearest impact point of current location, this impact point can be outside effective sensing range of sensor.

Usually when global path planning, impact point is described under navigation coordinate system, as earth inertial coodinate system, sky, northeast coordinate system etc.; For ease of finishing local paths planning, set up the local paths planning coordinate system, this coordinate is an instantaneous coordinate system, the origin of coordinates be positioned at when preplanning mobile robot's barycenter constantly on the subpoint on the horizontal plane, X-axis is pointed to mobile robot's direction of advance, Y-axis is positioned on the horizontal plane, points to the mobile robot left side, and the Z axle satisfies right-hand screw rule.

The definition difference of navigation coordinate system, then the transformational relation that is tied to the local paths planning coordinate system from navigation coordinate is not quite similar, therefore the present invention no longer provides concrete conversion formula, and in the follow-up local paths planning step, all calculating all will be finished under the local paths planning coordinate system.

In the local paths planning process, according to the environment basic assumption, landform is comparatively smooth, can be reduced to two-dimensional space, and therefore, the X and the Y direction coordinate that only need to calculate impact point get final product, and the Z coordinate can not considered.

The definition of local paths planning coordinate system as shown in Figure 3, the below rectangle represent mobile robot's outline among the figure, suppose that the point 1 that the upper right corner indicates is an impact point, then can obtain this coordinate on local paths planning coordinate system X-axis and Y direction.

Acquiescence path planning when (2) calculating is accessible.

According to the coordinate (be impact point and mobile robot's relative position relation) of impact point under the local paths planning coordinate system, acquiescence output path planning that can be when accessible.

As shown in Figure 4, acquiescence output path planning was positioned at the initial point and the impact point of local paths planning coordinate system, and with the tangent circle of X-axis on.If impact point is positioned on the X-axis, then acquiescence output path planning deteriorates to line segment from segmental arc.The curvature of this path planning is curvature that this circle is corresponding, and satisfy turn left for just, turn right to bearing.

If acquiescence output curvature is S ₀, the impact point coordinate is (S ₀, y ₀), then have: ${\mathrm{<figure-callout\; id="0"\; label="S"\; filenames="A20071012319800331.png"\; state="\{\{state\}\}">S</figure-callout>}}_0=2{\mathrm{<figure-callout\; id="0"\; label="y"\; filenames="A20071012319800331.png"\; state="\{\{state\}\}">y</figure-callout>}}_0/(x_0^2+{\mathrm{<figure-callout\; id="0"\; label="y"\; filenames="A20071012319800331.png"\; state="\{\{state\}\}">y</figure-callout>}}_0^2).$

Acquiescence output arc length L ₀Selection need to take all factors into consideration according to actual conditions, should satisfy following several condition:

(a) as previously mentioned, the output arc length of local paths planning should be greater than 2Vt;

(b) acquiescence output arc length should be not more than sensor environment sensing scope, that is, the security of the corresponding segmental arc of acquiescence output arc length can the abundant perception by the data of current sensor, as, if the available field of view of camera distance is 5m, then acquiescence output arc length also should be not more than 5m;

(c) acquiescence output arc length is unsuitable excessive, is separated by one to give tacit consent to the environmental data that is obtained on two location points of output arc length distance and should have bigger coincidence, to hold environmental information as far as possible exactly.

If select the fixed cycle local paths planning, acquiescence output arc length is final output arc length, all no longer the output arc length is adjusted in each step of back; If select the variable period local paths planning, then acquiescence output arc length is actually minimum output arc length, may increase the output valve of planning arc length in the step (7) according to complaint message.

(3) initialization of local paths planning intermediate variable and parameter

For finishing path planning, the present invention has defined the left and right sides and has kept away the barrier segmental arc, the barrier segmental arc is kept away on a left side, and () segmental arc promptly the most on the right side, the barrier segmental arc was kept away for kept away the segmental arc in the required maximum curvature of current obstacle (i.e. the most close left side) from the right side in the right side in order to keep away the required minimum curvature of obstacle on the current impact point direction from the left side.As shown in Figure 5, for the obstacle among the figure, segmental arc 1 is a left side and keeps away the barrier segmental arc, and segmental arc 2 is the right side and keeps away the barrier segmental arc.About keep away the barrier segmental arc arc length L _l, L _rWith curvature S _l, S _rEqual dynamic calculation in step (7), the local optimum planning segmental arc of final output is also chosen from these two segmental arcs.Under the primary condition, two segmental arcs all overlap with the acquiescence planning segmental arc that calculates in the step (2), and S is promptly arranged _l=S _r=S ₀, L _l=L _r=L ₀

In the local paths planning subsequent step, also need to use following parameter: mobile robot's length a, width b, the min. turning radius R that the mobile robot can realize _MinOr maximum curvature S _Max, local paths planning coverage D.

R _MinAnd S _MaxBe scalar, and satisfy S _Max=1/R _Min, be the important parameter of mobile robot's travelling performance, for wheeled robot, be subjected to the restriction of steering wheel angle scope, if wish that turning to of each deflecting roller is all consistent with this instantaneous direction of advance of taking turns, then there is minimum of a value R in its radius of turn that can realize _Min, littler radius of turn need be turned by pivot turn or differential and be realized, and this need be avoided in the travelling continuously of mobile robot as far as possible usually, therefore the output segmental arc curvature of local paths planning need be limited in [S _Max, S _Max] in the interval.

Be subjected to the restriction of sensor perception, the scope of the local environmental information that the mobile robot obtains is limited, for accuracy and the security that guarantees local paths planning, environmental data beyond mobile robot's certain distance can think confidence level not high or to not basic influence of current motion, all can ignore, remember that this distance is local paths planning coverage D.Can choose D=min (D _s, D _t), D wherein _sBe the environment sensing coverage, can determine D according to sensor configuring condition and motion control precision _tBe the impact point distance

Though because generally can be chosen at impact point outside the environment sensing effective range, but at the motion latter end, impact point distance may be less than the environment sensing coverage, and this moment, the obstacle at impact point rear was not the required care of local paths planning.

(4), obtain the binary environmental information and describe, and will carry out preliminary classification the binary environmental information according to sensor output.

So-called binary environment supposes that promptly environment is made up of free space and obstacle, and in this planning algorithm, free space promptly refers to flat road surface, and the mobile robot can realize in free space that straight line or camber line advance; Obstacle then comprises the space that all mobile robots can't pass through thereon, as has stone, hole, abrupt slope of certain volume etc.

Adopt the rectangular area to simplify in the local paths planning method of the present invention and describe complaint message, each limit of rectangle is parallel to the X-axis and the Y-axis of local paths planning coordinate system respectively, as shown in Figure 3.Can regard as for erose obstacle is that a plurality of rectangle obstacles overlap to form.Therefore for describing current binary environmental information, four coordinates of summit under the local paths planning coordinate system only need noting all obstacles get final product, and promptly placing obstacles to hinder adds up to n, and environmental information is described as the left front point coordinates (x of each point _Lfi, y _Lfi), right front point coordinates (x _Rfi, y _Rfi), left back point coordinates (x _Lri, y _Lri), right back point coordinates (x _Rri, y _Rri), wherein, i=1,2 ..., n.Can further be reduced to the front and back X coordinate and the left and right sides Y coordinate (x that adopt obstacle _Fi, x _Ri, y _Li, y _Ri) describe, as shown in Figure 3.

According to the environment basic assumption, in the local paths planning process, needn't consider complex-terrain and on a large scale situation such as large tracts of land obstacle to the influence of planning, this environment describing method has carried out maximum simplification to environment, thereby can reduce the requirement to the sensor hardware configuration.Usually, under, the rational prerequisite in installation site obvious, only depend on a camera, handle, can obtain to satisfy the environment descriptor of this local path planning algorithm through suitable visual pattern at environmental characteristic.The concrete generative process of environmental information is different because of the difference of sensor configuration, as long as but satisfy aforesaid data mode, just can be used for this local path planning algorithm, so the specific implementation that environmental data generates is not within the limit of consideration of this algorithm.

In this local paths planning method of invention, for the ease of management, all obstacles have all been added key words sorting, it is divided into different classifications, and in subsequent step, carries out different processing.According to obstacle with about keep away the relative position relation of barrier segmental arc, obstacle can be divided into four classes, promptly unknown obstacle, left side obstacle, right side obstacle, can ignore obstacle.Unknown obstacle be obstacle with about keep away relative position relation the unknown of barrier segmental arc; The left side obstacle is that obstacle is positioned at the right side and keeps away barrier segmental arc left side; The right side obstacle is that obstacle is positioned at a left side and keeps away barrier segmental arc right side; Can ignore obstacle and comprise two class situations, one be obstacle about keep away between the barrier segmental arc, then can think that this obstacle was kept away this moment, can not consider, its two for obstacle to mobile robot's distance greater than coverage D, can not influence the execution of this route programming result, therefore in current local paths planning process, can not consider.Concrete obstacle and segmental arc position relation judge and classification will be finished in step (6), and under the primary condition, all obstacles all belong to unknown obstacle.

Step (1) is the preparatory stage of local paths planning method to (4).

(5) choose current optimum segmental arc.

Keep away the barrier segmental arc relatively (as previously mentioned, two segmental arcs overlap under the original state, but in the step (7) of back after dynamic calculation assignment again, both promptly may be no longer identical), immediate from wherein choosing curvature with the curvature of acquiescence path planning, be considered as current optimum segmental arc, remember that its curvature and arc length are S _cAnd L _cPromptly compare | S _l-S ₀| and | S ₀-S _r|, if | S _l-S ₀|≤| S ₀-S _r|, then make S _c=S _l, L _c=L _l, otherwise make S _c=S _r, L _c=L _r

A precondition that it should be noted that above-mentioned comparison is that a left side is kept away the barrier segmental arc and right kept away the min. turning radius restriction that the barrier segmental arc does not all exceed the mobile robot, promptly max (| S _l|, | S _r|)≤S _MaxIf about keep away barrier and have one to exceed this restriction in the segmental arc, then directly choosing the segmental arc that does not transfinite is current optimum segmental arc, if two segmental arcs all transfinite, mean that then the obstacle in the current environment has blocked all directions of advance of mobile robot, therefore directly interrupt current local paths planning process, and return error message.

(6) relative position relation of dyscalculia and current optimum segmental arc and obstacle reclassified.

As previously mentioned, in this algorithm obstacle is divided into four classes, promptly unknown obstacle, left side obstacle, right side obstacle, can ignores obstacle.Definition according to the left side obstacle, the left side obstacle is positioned at the left side that the barrier segmental arc is kept away on the right side, be updated to unidirectional search to the right and the barrier segmental arc is kept away on the right side in step (7), be the new right side keep away the barrier segmental arc only may appear at the former right side keep away the barrier segmental arc the right side, in case determine that certain obstacle is the left side obstacle, whether no matter the barrier segmental arc is kept away on the right side renewal has taken place in planning process, and this obstacle all is well-determined with right relative position direction of keeping away the barrier segmental arc, is positioned at the left side that the barrier segmental arc is kept away on the right side all the time.Therefore, be that barrier is kept away during segmental arc in the right side in current optimum segmental arc, need not to recomputate the relative position relation of left side obstacle and current optimum segmental arc.In like manner, be that barrier is kept away during segmental arc in a left side in current optimum segmental arc, need not to recomputate the relative position relation of right side obstacle and current optimum segmental arc.According to the definition that can ignore obstacle, this obstacle is positioned at the left and right sides all the time to be kept away outside barrier segmental arc centre or the planning coverage, need not to calculate the relative position relation that can ignore obstacle and current optimum segmental arc equally.

To sum up, current optimum segmental arc is that the relative position relation of only more unknown obstacle and left side obstacle and segmental arc is kept away when hindering segmental arc on a left side; Current optimum segmental arc is that the relative position relation of only more unknown obstacle and right side obstacle and segmental arc is kept away when hindering segmental arc on the right side.

Order reads the information of obstacle.If unknown obstacle and obstacle and mobile robot's distance then directly can be ignored obstacle with its heavy being categorized as greater than the planning coverage; Otherwise the relative position relation of dyscalculia and current optimum segmental arc, disturbance in judgement be on the left side or the right side of optimum segmental arc, or drop on the optimum segmental arc.If obstacle is on optimum segmental arc, then interrupt step (6) enters step (7); Otherwise obstacle is reclassified, read next complaint message then, finish if whole complaint message has all read, still not because obstacle on segmental arc and interrupt, do not have obstacle on the then current optimum segmental arc, be feasible planning segmental arc, can directly change step (8) output program results over to.

The realization of this step can specifically be decomposed as follows:

(a) order reads complaint message (x _f, x _r, y _l, y _r), if there is no obstacle, or whole complaint message has all read and has finished, and then current optimum segmental arc is feasible planning segmental arc, can directly change step (8) output program results over to.

I. current optimum segmental arc is that the barrier segmental arc is kept away on a left side:

Maybe can ignore obstacle if obstacle is the right side obstacle, re-execute (a) and read next complaint message;

If obstacle is unknown obstacle or left side obstacle, change (b) over to.

Ii. current optimum segmental arc is that the barrier segmental arc is kept away on the right side:

Maybe can ignore obstacle if obstacle is the left side obstacle, re-execute (a) and read next complaint message;

If obstacle is unknown obstacle or right side obstacle, change (b) over to.

(b) if obstacle is unknown obstacle, judge its with mobile robot's distance whether greater than planning coverage D, if greater than D, then directly its heavy being categorized as can be ignored obstacle, return (a) then and read next complaint message.

The boundary coordinate of known obstacle is (x _f, x _r, y _l, y _r), represent the Y coordinate on front and rear side X coordinate and limit, the left and right sides respectively, the curvature of segmental arc and arc length are S _cAnd L _c, mobile robot's length is that a, width are b, the planning coverage is D.Judge whether unknown obstacle is planning that the formula outside the coverage is:

If y _lY _r≤ 0, then obstacle is on X-axis, and obstacle is to mobile robot's distance D _Ob=x _r

If y _r＞0, then obstacle is in the left side of robot, $D_{\mathrm{ob}}=\min (x_r,\sqrt{{(x_r-a/2)}^2+{(y_r-b/2)}^2});$

If y _l＜0, then obstacle is on the right side of robot, $D_{\mathrm{ob}}=\min (x_r,\sqrt{{(x_r-a/2)}^2+{(y_l+b/2)}^2}).$

If D _Ob＞D, then obstacle is for can ignore obstacle.

(c) according to calculate judging that the position of the current obstacle that reads with current optimum segmental arc concerns, promptly obstacle be left side, segmental arc in segmental arc the right side, still drop on the segmental arc.

As shown in Figure 6, because the mobile robot has its volume, so robot is not a camber line but one section annulus in the zone that is covered when current optimum segmental arc is advanced, and the center of circle of ring and outer shroud all overlaps with the center of circle of current optimum segmental arc in the annulus.Known current optimum segmental arc is S _c, then the segmental arc radius is R _c=1/S _c, central coordinate of circle be (0, R _c).The internal diameter of moving region annulus is | R _c|-b/2, external diameter is

(perhaps the value of external diameter can be similar to and be reduced to | R _c|+b/2).If any point of obstacle is dropped on and all can be influenced the mobile robot in this annular region and advance along this segmental arc, should regard obstacle as on segmental arc.

Can step (c) be divided into three kinds of situations according to the current optimum segmental arc of robot:

I. current optimum segmental arc is a straightway

Along with S _cLevel off to 0, R _c=1/S _cBe tending towards infinitely great, therefore work as | S _c|≤δ _cThe time, can think that mobile robot's walking path is straightway, wherein δ _cIt is an a small amount of.Have this moment:

If y _r〉=b/2, obstacle is in the segmental arc left side;

If y _l≤-b/2, obstacle is on the segmental arc right side;

If above-mentioned two are not satisfied, obstacle is on segmental arc.

Ii. current optimum segmental arc is the segmental arc of turning left (be the segmental arc center of circle+Y-axis on)

Work as S _c＞δ _cThe time, the center of circle is on+Y-axis.Ring was with interior (the right front point of obstacle to distance of center circle from less than internal diameter) in obstacle dropped on fully, and then obstacle is on the left of segmental arc; Obstacle drops on beyond the outer shroud (the left back point of obstacle to distance of center circle from greater than external diameter) fully, and then obstacle is on the segmental arc right side; Otherwise obstacle is on segmental arc.That is:

If $\sqrt{{\left(x_f\right)}^2+{(y_r-R_c)}^2}\&le;\left|R_c\right|-b/2$ , obstacle is in the segmental arc left side;

If $\sqrt{{\left(x_r\right)}^2+{(y_l-R_c)}^2}\&GreaterEqual;\sqrt{{\left(\right|R_c|+b/2)}^2+{(a/2)}^2}$ , obstacle is on the segmental arc right side;

If above-mentioned two are not satisfied, obstacle is on segmental arc.

Iii. current optimum segmental arc is the segmental arc of turning right (be the segmental arc center of circle-Y-axis on)

Work as S _c＜-δ _cThe time, the center of circle is on-Y-axis.Obstacle drops on beyond the outer shroud (the right back point of obstacle to distance of center circle from greater than external diameter) fully, and then obstacle is in the segmental arc left side; Ring was with interior (the left front point of obstacle to distance of center circle from less than internal diameter) in obstacle dropped on fully, and then obstacle is on the segmental arc right side; Otherwise obstacle is on segmental arc.That is:

If $\sqrt{{\left(x_r\right)}^2+{(y_r-R_c)}^2}\&GreaterEqual;\sqrt{{\left(\right|R_c|+b/2)}^2+{(a/2)}^2}$ , obstacle is in the segmental arc left side;

If $\sqrt{{\left(x_f\right)}^2+{(y_l-R_c)}^2}\&le;\left|R_c\right|-b/2$ , obstacle is on the segmental arc right side;

If above-mentioned two are not satisfied, obstacle is on segmental arc.

(d) according to the result of calculation of (c), if obstacle on current optimum segmental arc, then current optimum segmental arc is infeasible, interrupt step (6) enters keeps away the barrier segmental arc about step (7) recomputates and chooses; Otherwise (obstacle is in the left side of segmental arc or on the right side of segmental arc) enters (e).

(e) be that the barrier segmental arc is kept away on a left side or barrier segmental arc, the original affiliated classification of current obstacle and result of calculation (c) are kept away in the right side according to current optimum segmental arc, current obstacle is reclassified.

I. current optimum segmental arc is that the barrier segmental arc is kept away on a left side:

For unknown obstacle, if obstacle is kept away barrier segmental arc left side on a left side, then it must also keep away left side of barrier segmental arc on the right side, so heavily be categorized as the left side obstacle; If obstacle is kept away barrier segmental arc right side on a left side, then belong to the right side obstacle by the definition nature.

For the left side obstacle, if obstacle is kept away barrier segmental arc left side on a left side, classification does not change; If obstacle is kept away the right side of barrier segmental arc on a left side, according to class definition, obstacle is also kept away the left side of barrier segmental arc simultaneously on the right side, therefore heavily is categorized as and can ignores obstacle.

Ii. current optimum segmental arc is that the barrier segmental arc is kept away on the right side:

For unknown obstacle,, then belong to the left side obstacle by the definition nature if obstacle is kept away barrier segmental arc left side on the right side; If obstacle is kept away barrier segmental arc right side on the right side, then it must also keep away right side of barrier segmental arc on a left side, so heavily be categorized as the right side obstacle.

For the right side obstacle, if obstacle is kept away barrier segmental arc left side on the right side, according to class definition, obstacle is also kept away the right side of barrier segmental arc simultaneously on a left side, therefore heavily is categorized as and can ignores obstacle; If obstacle is kept away the right side of barrier segmental arc on the right side, classification does not change.

Step (e) can be concluded as table 1.

The classification of table 1 obstacle

Current optimum segmental arc	The barrier segmental arc is kept away on a left side		The barrier segmental arc is kept away on the right side
					The former classification of obstacle	Unknown obstacle	The left side obstacle	Unknown obstacle	The right side obstacle
Obstacle is in the stylish classification in segmental arc left side	The left side obstacle	The left side obstacle	The left side obstacle	Can ignore obstacle
					Obstacle is in the stylish classification in segmental arc right side	The right side obstacle	Can ignore obstacle	The right side obstacle	The right side obstacle

(f) return (a), read next complaint message.

(7) keep away the barrier segmental arc about the renewal

When result of calculation showed that certain obstacle is on current optimum segmental arc, step (6) was interrupted, and enters step (7), kept away the barrier segmental arc about upgrading according to this complaint message.

If being a left side, current optimum segmental arc keeps away the barrier segmental arc, it is invalid to prove that the barrier segmental arc is kept away on a former left side, need to continue to seek a new left side to the left and keeps away the barrier segmental arc, promptly calculates and keeps away current obstacle left side edge point required minimum curvature and arc length, and assignment as a result kept away the barrier segmental arc to a left side, the right side is kept away the barrier segmental arc and is remained unchanged; If being the right side, current optimum segmental arc keeps away the barrier segmental arc, it is invalid to prove that the barrier segmental arc is kept away on the former right side, need to continue to seek the new right side to the right and keeps away the barrier segmental arc, promptly calculates and keeps away current obstacle right side edge point required maximum curvature and arc length, and assignment as a result kept away the barrier segmental arc to the right side, a left side is kept away the barrier segmental arc and is remained unchanged.

That is, current optimum segmental arc is that a left side is kept away when hindering segmental arc, then upgrades a left side and keeps away the barrier segmental arc; Current optimum segmental arc is that the right side is kept away when hindering segmental arc, then upgrades the right side and keeps away the barrier segmental arc.

(a) upgrade a left side and keep away the barrier segmental arc

The known boundary coordinate that drops on the current obstacle on the current optimum segmental arc is (x _f, x _r, y _l, y _r), mobile robot's length is that a, width are b, the planning coverage is D, requires an obstacle and a left side to keep away the barrier segmental arc and keeps $d=\sqrt{{(a/2)}^2+{(b/2)}^2}$ Safe distance, a left side keep away the barrier segmental arc calculating as shown in Figure 7.Provided two kinds of different situations among the figure, for the obstacle shown in the left side, the segmental arc that the barrier segmental arc should be a left-hand bend is kept away on its left side;

For the obstacle shown in the right side, the segmental arc that the barrier segmental arc should be a right-hand bend is kept away on its left side.Two kinds of situations can be distinguished according to the coordinate of obstacle left margin.

I. the left barrier segmental arc of keeping away is a segmental arc of turning left

If home position be (0, R _l), when the left margin of obstacle satisfies y _l+ d 〉=0 has:

$\sqrt{{\left(x_r\right)}^2+{(y_l-R_l)}^2}=R_l+d$

$S_l=\frac{1}{R_l}=\frac{2(y_l+d)}{{\left(x_r\right)}^2+{\left(y_l\right)}^2-{\mathrm{<figure-callout\; id="2"\; label="d"\; filenames="A20071012319800321.png,A20071012319800333.png"\; state="\{\{state\}\}">d</figure-callout>}}^2}$

Ii. a left side keep away the barrier segmental arc be a segmental arc that bends to right

When the left margin of obstacle satisfies y _l+ d＜0 has:

$\sqrt{{\left(x_f\right)}^2+{(y_l-R_l)}^2}=|R_l+d|$

$S_l=\frac{1}{R_l}=\frac{2(y_l+d)}{{\left(x_f\right)}^2+{\left(y_l\right)}^2-{\mathrm{<figure-callout\; id="2"\; label="d"\; filenames="A20071012319800321.png,A20071012319800333.png"\; state="\{\{state\}\}">d</figure-callout>}}^2}$

Above-mentioned two S _lComputing formula can unify be:

$S_l=\frac{1}{R_l}=\frac{2(y_l+d)}{{\left(x_i\right)}^2+{\left(y_l\right)}^2-{\mathrm{<figure-callout\; id="2"\; label="d"\; filenames="A20071012319800321.png,A20071012319800333.png"\; state="\{\{state\}\}">d</figure-callout>}}^2}$

Work as y in the formula _l+ d 〉=0 o'clock x _i=x _r, work as y _l+ d＜0 o'clock x _i=x _f

If the local paths planning algorithm adopts variable period to carry out, the arc length that then also needs a left side to be kept away the barrier segmental arc is upgraded.The arc length computing formula is:

$L_l=\min [D,\max (L_l,|R_l\mathrm{si}{n}^{-1}\frac{x_f}{R_l+d}\left|\right)]$

(b) upgrade the right side and keep away the barrier segmental arc

In like manner can get the right more new formula of keeping away the barrier segmental arc is:

$S_r=\frac{1}{R_r}=\frac{2(y_r-d)}{{\left(x_i\right)}^2+{\left(y_r\right)}^2-{\mathrm{<figure-callout\; id="2"\; label="d"\; filenames="A20071012319800321.png,A20071012319800333.png"\; state="\{\{state\}\}">d</figure-callout>}}^2}$

Work as y in the formula _r-d 〉=0 o'clock x _i=x _f, work as y _r-d＜0 o'clock x _i=x _rCorresponding arc length computing formula is:

$L_r=\min [D,\max (L_r,|R_r\mathrm{si}{n}^{-1}\frac{x_f}{R_r-d}\left|\right)]$

(c) treatment of special situation

The curvature S that is calculating _lOr S _rLevel off to zero the time, the radius infinity can appear in calculating, and can change top arc length computing formula into this moment:

L＝min[D，max(L，x _f)]

Keep away about upgrading after the barrier segmental arc, obstacle no longer drops on the segmental arc, keeps away the barrier segmental arc for a left side, obstacle is kept away the barrier segmental arc on the segmental arc right side for the right side at this moment, and this moment, obstacle was in the segmental arc left side, so can be according to the category attribute of obstacle, by the rule of table 1 current obstacle is reclassified.

After finishing above-mentioned work, return step (5) again relatively and choose current optimum segmental arc, enter circulation next time.

(8) local paths planning result output

If behind all obstacles of traversal, do not have obstacle to drop on the current optimum segmental arc, then current optimum segmental arc is feasible planning segmental arc, with curvature S in the step (6) _cWith arc length L _cExport to navigation and control system as current optimum programming path, and finish current local paths planning process.

Navigation is called the local paths planning algorithm with control system every some cycles, obtain the local optimum path planning, and the controlled motion system advances along this path, through several all after dates can be progressively near so that the final range of allowable error that arrives impact point within, as shown in Figure 8.

As can be seen from the above step, the realization of local paths planning method of the present invention is very simple, basically rely on how much calculating can finish complete planning computational process fully, the planning time can finish in a short period of time, thereby satisfies the planning real-time requirement of mobile robot's continuous walking; For snagged environment, in step (6), all do not travel through all obstacles at every current optimal path, but in case find that certain obstacle drops on current segmental arc and makes a response at once, again compare the relation of obstacle and segmental arc again after keeping away the barrier segmental arc about renewal, thereby can greatly improve executing efficiency; Whole planning algorithm flow process is simply distinct, should be readily appreciated that easily engineering programming; The path planning of this local path planning algorithm output is and the tangent segmental arc of current mobile robot's direction of advance, thereby makes mobile robot's entire motion track can form the curve of a continuous slyness, is convenient to the realization of motion control.

Inspection tour prober for moon surface principle prototype with Chinese Academy of Space Technology's independent research is that example further specifies specific implementation process of the present invention:

This principle prototype is imaginary background with the lunar surface environment.Known lunar surface environment is comparatively smooth mellow soil, and the obstacle of certain probability that distributes, and satisfies the basic assumption of this local path planning algorithm.This principle prototype is six to take turns rocker arm suspension formula structure, six wheel drive, four horn rings in front and back provide and turn to, can realize advancing, retreat, pivot turn, advance in Ackerman turn.The main sensors that is used for local paths planning be a pair of be installed on car body the place ahead keep away the barrier camera.

The min. turning radius that requires in the model machine motion-control module is 1.5 meters, and the length-width ratio of model machine is 1000mm * 800mm, and the travel speed of model machine is about 20～25mm/s when independently walking.Effective visual range of keeping away the barrier camera is about 3～5 meters, keep away barrier camera total angle of visual field that can provide for two and be about 100 degree, and do not provide the function that keeps the history environment data on the car, therefore select the local paths planning algorithm of variable period pattern for use, promptly, output planning arc length L is a variate, and sets model machine and started the local paths planning process from last time, starts local paths planning next time when operating range reaches 0.8L.At the situation of principle prototype, this local path planning algorithm partial content is further simplified simultaneously.

Being implemented as follows of local paths planning method of the present invention:

(1) calculates the coordinate of impact point under the local paths planning coordinate system.

The navigation coordinate of inspection tour prober for moon surface principle prototype is selenographic coordinates systems, its XOY plane is parallel with the XOY plane of local paths planning coordinate system, therefore Coordinate Conversion is that simple horizontal translation adds the rotation of Z axle, this Coordinate Conversion is finished calculating by navigation and control system, and imports in the lump when calling the local paths planning algorithm.

Acquiescence path planning when (2) calculating is accessible.

If acquiescence output curvature is S ₀, the impact point coordinate is (x ₀, y ₀), have:

${\mathrm{<figure-callout\; id="0"\; label="S"\; filenames="A20071012319800331.png"\; state="\{\{state\}\}">S</figure-callout>}}_0=2{\mathrm{<figure-callout\; id="0"\; label="y"\; filenames="A20071012319800331.png"\; state="\{\{state\}\}">y</figure-callout>}}_0/({\mathrm{<figure-callout\; id="0"\; label="x"\; filenames="A20071012319800331.png"\; state="\{\{state\}\}">x</figure-callout>}}_0^2+{\mathrm{<figure-callout\; id="0"\; label="y"\; filenames="A20071012319800331.png"\; state="\{\{state\}\}">y</figure-callout>}}_0^2)$

The translational speed of principle prototype is lower, calculates 2Vt less than 0.1m, but because in the autonomous walking process, distant end may not have detailed environmental information feedback, requires the interval of local paths planning under the accessible situation unsuitable long, therefore chooses L ₀Be 0.5m.

(3) initialization of local paths planning algorithm intermediate variable and parameter.

About keep away the barrier segmental arc arc length L _l, L _rWith curvature S _l, S _rGet S under the primary condition _l=S _r=S ₀, L _l=L _r=L ₀

The length a=1 of sampler, width b=0.8, the min. turning radius R that the mobile robot can realize _Min=1.5 is maximum curvature S _Max=0.667, local paths planning environment sensing coverage D _s=3.The output segmental arc curvature of local paths planning is limited in [0.667,0.667] interval.

The local paths planning coverage $D=\min (D_s,\sqrt{x_0^2+{\mathrm{<figure-callout\; id="0"\; label="y"\; filenames="A20071012319800331.png"\; state="\{\{state\}\}">y</figure-callout>}}_0^2}).$

(4), obtain the binary environmental information and describe, and carry out the initial obstacle classification according to sensor output.

The vision module of principle prototype is finished the collection and the processing of camera photographic images, and according to the data of local paths planning algorithm input requirement, with the complaint message in the environment with (x _Fl, x _Ri, y _Li, y _Ri) form offer the local paths planning algorithm, i=1 wherein, 2 ..., n, n are the obstacle sum, x _Fi, x _Ri, y _Li, y _RiThe preceding border X coordinate of respectively corresponding each obstacle, border, back X coordinate, left margin Y coordinate, right margin Y coordinate.Under the primary condition, all obstacles all are categorized as unknown obstacle.

(5) choose current optimum segmental arc.

Curvature and the arc length of remembering current optimum segmental arc are S _cAnd L _c, keep away the barrier segmental arc about comparing, if | S _l-S ₀|≤| S ₀-S _r|, then make S _c=S _l, L _c=L _l, otherwise make S _c=S _r, L _c=L _r

(6) relative position relation of dyscalculia and current optimum segmental arc and obstacle reclassified.

Order reads complaint message (x _f, x _r, y _l, y _r), if unknown obstacle judges whether outside the planning coverage:

If y _lY _r≤ 0, then obstacle is on X-axis, and obstacle is to the distance D of principle prototype _Ob=x _r

If y _r＞0, then obstacle is in the left side of model machine, $D_{\mathrm{ob}}=\min (x_r,\sqrt{{(x_r-a/2)}^2+{(y_r-b/2)}^2});$

If y _l＜0, then obstacle is on the right side of model machine, $D_{\mathrm{ob}}=\min (x_r,\sqrt{{(x_r-a/2)}^2+{(y_l+b/2)}^2}).$

If D _Ob＞D, then obstacle is for can ignore obstacle.

Because the running environment of principle prototype is soft lunar soil, has certain slippage, positioning accuracy is relatively limited, and the gabarit of the covering annular section that model machine is advanced requires can be very accurate, therefore is reduced to the external diameter of moving region annulus is approximate | R _c|+b/2, internal diameter still be | R _c|-b/2.If any point of obstacle is dropped on and all can be influenced the mobile robot in this annular region and advance along this segmental arc, should regard obstacle as on segmental arc.

Get δ _c=0.01, when | S _c|≤δ _cThe time, can think that the walking path of model machine is a straightway, have this moment:

If y _r〉=b/2, obstacle is in the segmental arc left side;

If y _l≤-b/2, obstacle is on the segmental arc right side;

If above-mentioned two are not satisfied, obstacle is on segmental arc.

Work as S _c＞δ _cThe time, the center of circle is on+Y-axis:

If $\sqrt{{\left(x_f\right)}^2+{(y_r-R_c)}^2}\&le;\left|R_c\right|-b/2$ , obstacle is in the segmental arc left side;

If $\sqrt{{\left(x_r\right)}^2+{(y_l-R_c)}^2}\&GreaterEqual;\left|R_c\right|+b/2$ , obstacle is on the segmental arc right side;

If above-mentioned two are not satisfied, obstacle is on segmental arc.

Work as S _c＜-δ _cThe time, the center of circle is on-Y-axis:

If $\sqrt{{\left(x_r\right)}^2+{(y_r-R_c)}^2}\&GreaterEqual;\left|R_c\right|+b/2$ , obstacle is in the segmental arc left side;

If $\sqrt{{\left(x_f\right)}^2+{(y_l-R_c)}^2}\&le;\left|R_c\right|-b/2$ , obstacle is on the segmental arc right side;

If above-mentioned two are not satisfied, obstacle is on segmental arc.

If obstacle is on segmental arc, by the principle of table 1 obstacle is reclassified, if just drop on the segmental arc, then current optimum segmental arc is infeasible, needs interrupt step (6) to enter and keeps away the barrier segmental arc about step (7) recomputates and chooses.Perhaps, behind all obstacles of traversal, do not have obstacle to drop on the current optimum segmental arc, then current optimum segmental arc is feasible planning segmental arc, can directly change step (8) output program results over to.

(7) keep away the barrier segmental arc about the renewal.

As previously mentioned, based on the consideration of the sliding phenomenon and the positioning accuracy of model machine, obstacle is chosen at the safe distance of keeping away the barrier segmental arc

The basis on slightly relax, get d=0.8.A left side keep away the barrier segmental arc calculating as shown in Figure 7.Note curvature computing formula is:

$S_m=\frac{2(y_m\&PlusMinus;d)}{{\left(x_n\right)}^2+{\left(y_m\right)}^2-{\mathrm{<figure-callout\; id="2"\; label="d"\; filenames="A20071012319800321.png,A20071012319800333.png"\; state="\{\{state\}\}">d</figure-callout>}}^2}$

Work as y _m± d 〉=0 o'clock x _n=x _f, otherwise x _n=x _r

An above corresponding left side is kept away when hindering segmental arc subscript _mReplace with _l, ± get just; The corresponding right side is kept away when hindering segmental arc subscript _mReplace with _r, ± get is negative.

The arc length computing formula is reduced to:

L＝min[D，max(L，x _f)]

Keep away the barrier segmental arc if current optimum segmental arc is a left side, then only upgrade a left side and keep away the barrier segmental arc, the right side is kept away the barrier segmental arc and is remained unchanged; Keep away the barrier segmental arc if current optimum segmental arc is the right side, then only upgrade the right side and keep away the barrier segmental arc, a left side is kept away the barrier segmental arc and is remained unchanged.

Keep away about upgrading after the barrier segmental arc, obstacle no longer drops on the segmental arc, keeps away the barrier segmental arc for a left side, obstacle is kept away the barrier segmental arc on the segmental arc right side for the right side at this moment, and this moment, obstacle was in the segmental arc left side, so can be according to the category attribute of obstacle, by the rule of table 1 current obstacle is reclassified.

After finishing above-mentioned work, return step (4) again relatively and choose current optimum segmental arc, enter circulation next time.

(8) local paths planning result output.

If behind all obstacles of traversal, do not have obstacle to drop on the current optimum segmental arc, then current optimum segmental arc is feasible planning segmental arc, with curvature S in the step (6) _cWith arc length L _cExport to navigation and control system as current optimum programming path, and finish current local paths planning process.

Above-mentioned steps is consistent with previously described local paths planning method basically, only done small adjustment in detail, emulation and interior outfield experiments have all proved, when the distance of obstacle model machine is crossed near or distance objective was put near, because the restriction of principle prototype sensor environment acquisition field of view and kinetic characteristic, can't realize that the autonomous barrier of keeping away moves, other generally speaking, local paths planning by continuous several times, the detector principle prototype all can be according to the output of planning algorithm, and near-earth arrived impact point as far as possible when keeping away obstacle.

Claims (12)

1, a kind of mobile robot's local paths planning method based on the binary environmental information is characterized in that step is as follows:

(1) calculates the coordinate x of impact point under the local paths planning coordinate system ₀, y ₀

Acquiescence path planning when (2) calculating is accessible comprises that acquiescence output curvature is S ₀With acquiescence output arc length L ₀

(3) intermediate variable, promptly barrier segmental arc and local paths planning parameter initialization are kept away in the left and right sides

The barrier segmental arc is kept away for kept away the required minimum curvature of obstacle on the current impact point direction from the left side in a definition left side, i.e. segmental arc the most on the right side, the barrier segmental arc is kept away for kept away the required maximum curvature of current obstacle from the right side in the right side, i.e. the segmental arc in close left side, about keep away the barrier segmental arc arc length be respectively L _l, L _r, about keep away the barrier segmental arc curvature be respectively S _l, S _r, under the primary condition, two segmental arcs all overlap with the acquiescence planning segmental arc that calculates in the step (2), and S is promptly arranged _l=S _r=S ₀, L _l=L _r=L ₀

The parameter of local paths planning is: mobile robot's length a, width b, the min. turning radius R that the mobile robot can realize _MinOr maximum curvature S _Max, local paths planning coverage D, R _MinAnd S _MaxBe scalar, and satisfy S _Max=1/R _Min, the output segmental arc curvature of local paths planning is limited in [S _Max, S _Max] in the interval, D=min (D _s, D _t), D wherein _sBe the environment sensing coverage, can determine D according to sensor configuring condition and motion control precision _tBe the impact point distance

(4), obtain the binary environmental information and describe, and the complaint message in the binary environmental information is carried out preliminary classification according to sensor output

A. with the complaint message in the binary environment with x _Fi, x _Ri, y _Li, y _RiForm provide, the respectively preceding border X coordinate of corresponding each obstacle, border, back X coordinate, left margin Y coordinate, right margin Y coordinate, i=1,2 ..., n, n are the obstacle sum;

B. according to obstacle with about keep away the relative position relation of barrier segmental arc, obstacle is divided into four classes, promptly unknown obstacle, left side obstacle, right side obstacle, can ignore obstacle; Unknown obstacle be obstacle with about keep away relative position relation the unknown of barrier segmental arc; The left side obstacle is that obstacle is positioned at the right side and keeps away barrier segmental arc left side; The right side obstacle is that obstacle is positioned at a left side and keeps away barrier segmental arc right side; Can ignore obstacle and comprise two class situations, one be obstacle about keep away between the barrier segmental arc, then can think that this obstacle was kept away this moment, can not consider, its two for obstacle to mobile robot's distance greater than coverage, can not influence the execution of this route programming result, therefore in current local paths planning process, can not consider;

C. all obstacle preliminary classification are defined as unknown obstacle;

(5) choose current optimum segmental arc

Keep away the barrier segmental arc relatively, immediate from wherein choosing curvature with the curvature of acquiescence path planning, be considered as current optimum segmental arc, the curvature and the arc length of optimum segmental arc are respectively S _cAnd L _c, promptly compare | S _l-S ₀| and | S ₀-S _r|, if | S _l-S ₀|≤| S ₀-S _r|, then make S _c=S _l, L _c=L _l, otherwise make S _c=S _r, L _c=L _r

(6) relative position relation of the current optimum segmental arc of dyscalculia and step (5), and obstacle reclassified

Order reads the information of obstacle, if unknown obstacle and obstacle and mobile robot's distance then directly can be ignored obstacle with its heavy being categorized as greater than planning coverage D; Otherwise the relative position relation of dyscalculia and current optimum segmental arc, disturbance in judgement is on the left side or the right side of optimum segmental arc, or drop on the optimum segmental arc, if obstacle drops on the optimum segmental arc, then interrupt step (6) enters step (7), otherwise obstacle is reclassified, read next complaint message then; If all complaint messages have all read and have finished, still, be feasible planning segmental arc not because obstacle on segmental arc and interrupt, does not have obstacle on the then current optimum segmental arc, can directly change step (8) output program results over to;

(7) keep away the barrier segmental arc about the renewal

Current optimum segmental arc is that a left side is kept away when hindering segmental arc, then upgrades a left side and keeps away the barrier segmental arc; Current optimum segmental arc is that the right side is kept away when hindering segmental arc, then upgrade the right side and keep away the barrier segmental arc, about upgrading, keep away after the barrier segmental arc, obstacle no longer drops on the segmental arc, keeps away the barrier segmental arc for a left side, and this moment, obstacle was on the segmental arc right side, keep away the barrier segmental arc for the right side, this moment, obstacle was in segmental arc left side, and then according to the category attribute of obstacle, current obstacle reclassified;

After finishing above-mentioned work, return step (5) and choose optimum segmental arc again;

(8) output local paths planning result

If behind all obstacles of traversal, do not have obstacle to drop on the current optimum segmental arc, then current optimum segmental arc is feasible planning segmental arc, with curvature S in the step (6) _cWith arc length L _cExport to navigation and control system as current optimum programming path, and finish current local paths planning process.

2, the mobile robot's local paths planning method based on the binary environmental information according to claim 1, it is characterized in that: the origin of coordinates of the local paths planning coordinate system in the described step (1) be positioned at when preplanning mobile robot's barycenter constantly on the subpoint on the horizontal plane, X-axis is pointed to mobile robot's direction of advance, Y-axis is positioned on the horizontal plane, point to the mobile robot left side, the Z axle satisfies right-hand screw rule.

3, the mobile robot's local paths planning method based on the binary environmental information according to claim 1 is characterized in that: the acquiescence output curvature in the described step (2) $S_0={2y}_0/(x_0^2+y_0^2),$ (x ₀, y ₀) be the impact point coordinate.

4, the mobile robot's local paths planning method based on the binary environmental information according to claim 1 is characterized in that: the acquiescence output arc length L in the described step (2) ₀Selection principle:

(1) acquiescence of local paths planning output arc length L ₀Should be greater than 2Vt, wherein Vt is the distance that the mobile robot moves in the image processing time;

(2) acquiescence output arc length L ₀Should be not more than sensor environment sensing scope, i.e. acquiescence output arc length L ₀The security of corresponding segmental arc can the abundant perception by the data of current sensor;

(3) be separated by one and give tacit consent to output arc length L ₀The environmental data that is obtained on two location points of distance should have repeatability, so that hold environmental information exactly.

5, the mobile robot's local paths planning method based on the binary environmental information according to claim 1, it is characterized in that: described local paths planning adopts fixed cycle to carry out, or variable period is carried out two kinds of executive modes, for the fixed cycle local paths planning, acquiescence output arc length is final output arc length in the described step (2); For the variable period local paths planning, then acquiescence output arc length is minimum output arc length, may also can increase the output valve of planning arc length in the step (7) according to complaint message.

6, the mobile robot's local paths planning method based on the binary environmental information according to claim 1 is characterized in that: judge that whether unknown obstacle in the method outside the coverage D of planning is in the described step (6):

If y _lY _r≤ 0, then obstacle is on X-axis, and obstacle is to mobile robot's distance D _Ob=x _r

If y _r＞0, then obstacle is in the left side of robot, $D_{\mathrm{ob}}=\min (x_r,\sqrt{{(x_r-a/2)}^2+{(y_r-b/2)}^2});$

If y _l＜0, then obstacle is on the right side of robot, $D_{\mathrm{ob}}=\min (x_r,\sqrt{{(x_r-a/2)}^2+{(y_l+b/2)}^2})$

If D _Ob＞D, then obstacle be for can ignore obstacle,

(x _f, x _r, y _l, y _r) be the boundary coordinate of known obstacle, represent the Y coordinate on front and rear side X coordinate and limit, the left and right sides respectively, the curvature of segmental arc and arc length are S _cAnd L _c, a, b are respectively mobile robot's length and width.

7, the mobile robot's local paths planning method based on the binary environmental information according to claim 1, it is characterized in that: the relative position relation of described step (6) disturbance in judgement and segmental arc, be obstacle in the segmental arc left side, still drop on the segmental arc that its determination methods is on the segmental arc right side:

(1) as | S _c|≤δ _cThe time, mobile robot's walking path is a straightway, have this moment:

If y _r〉=b/2, obstacle is in the segmental arc left side;

If y _l≤-b/2, obstacle is on the segmental arc right side;

If above-mentioned two are not satisfied, obstacle is on segmental arc;

(2) work as S _c＞δ _cThe time, the center of circle is on+Y-axis:

If $\sqrt{{\left(x_f\right)}^2+{(y_r-R_c)}^2}\&le;\left|R_c\right|-b/2,$ Obstacle is in the segmental arc left side;

If $\sqrt{{\left(x_r\right)}^2+{(y_l-R_c)}^2}\&GreaterEqual;\left|R_c\right|+b/2,$ Obstacle is on the segmental arc right side;

If above-mentioned two are not satisfied, obstacle is on segmental arc;

(3) work as S _c＜-δ _cThe time, the center of circle is on-Y-axis:

If $\sqrt{{\left(x_r\right)}^2+{(y_r-R_c)}^2}\&GreaterEqual;\left|R_c\right|+b/2,$ Obstacle is in the segmental arc left side;

If $\sqrt{{\left(x_f\right)}^2+{(y_l-R_c)}^2}\&le;\left|R_c\right|-b/2,$ Obstacle is on the segmental arc right side;

If above-mentioned two are not satisfied, obstacle is on segmental arc;

δ wherein _cBe an a small amount of that sets in advance, (x _f, x _r, y _l, y _r) be the boundary coordinate of known obstacle, represent the Y coordinate on front and rear side X coordinate and limit, the left and right sides respectively, S _cAnd L _cBe respectively the curvature and the arc length of optimum segmental arc, a, b are respectively mobile robot's length and width.

8, the mobile robot's local paths planning method based on the binary environmental information according to claim 1, it is characterized in that: the obstacle principle of classification in the described step (6) is:

A. current optimum segmental arc is that a left side is kept away when hindering segmental arc:

For unknown obstacle, if obstacle is kept away barrier segmental arc left side on a left side, then it must also keep away left side of barrier segmental arc on the right side, so heavily be categorized as the left side obstacle; If obstacle is kept away barrier segmental arc right side on a left side, then belong to the right side obstacle by the definition nature;

For the left side obstacle, if obstacle is kept away barrier segmental arc left side on a left side, classification does not change; If obstacle is kept away the right side of barrier segmental arc on a left side, according to class definition, obstacle is also kept away the left side of barrier segmental arc simultaneously on the right side, therefore heavily is categorized as and can ignores obstacle.

B. current optimum segmental arc is that the right side is kept away when hindering segmental arc:

For unknown obstacle,, then belong to the left side obstacle by the definition nature if obstacle is kept away barrier segmental arc left side on the right side; If obstacle is kept away barrier segmental arc right side on the right side, then it must also keep away right side of barrier segmental arc on a left side, so heavily be categorized as the right side obstacle;

For the right side obstacle, if obstacle is kept away barrier segmental arc left side on the right side, according to class definition, obstacle is also kept away the right side of barrier segmental arc simultaneously on a left side, therefore heavily is categorized as and can ignores obstacle; If obstacle is kept away the right side of barrier segmental arc on the right side, classification does not change;

Mentioned above principle is reduced following table 1

Table 1 Current optimum segmental arc The barrier segmental arc is kept away on a left side The barrier segmental arc is kept away on the right side The former classification of obstacle Unknown obstacle The left side obstacle Unknown obstacle The right side obstacle Obstacle is in the stylish classification in segmental arc left side The left side obstacle The left side obstacle The left side obstacle Can ignore obstacle Obstacle is in the stylish classification in segmental arc right side The right side obstacle Can ignore obstacle The right side obstacle The right side obstacle 。

9, according to claim 1 or 8 described mobile robot's local paths planning methods based on the binary environmental information, it is characterized in that: keep away about described step (7) is upgraded after the barrier segmental arc again according to the category attribute of obstacle, reclassify according to above-mentioned table 1 pair current obstacle.

10, the mobile robot's local paths planning method based on the binary environmental information according to claim 1 is characterized in that: upgrading left computing formula of keeping away barrier segmental arc curvature in the described step (7) is:

$S_l=\frac{1}{R_l}=\frac{2(y_l+d)}{{\left(x_i\right)}^2+{\left(y_l\right)}^2-d^2}$

Work as y in the formula _l+ d 〉=0 o'clock x _i=x _r, work as y _l+ d＜0 o'clock x _i=x _f, a is a width for mobile robot's length, b, $d=\sqrt{{(a/2)}^2+{(b/2)}^2},$ R _lFor keeping away the Y coordinate that hinders the segmental arc center of circle in a left side, its absolute value is kept away the radius of barrier segmental arc to the center of circle for a left side, and D is for planning coverage, (x _f, x _r, y _l, y _r) be the boundary coordinate of known obstacle.

If adopt variable period to carry out in the local paths planning, the arc length that then also needs a left side to be kept away the barrier segmental arc is upgraded, and the arc length computing formula is:

$L_l=\min [D,\max (L_l,|R_l{\sin }^{-1}\frac{x_f}{R_l+d}\left|\right)].$

11, the mobile robot's local paths planning method based on the binary environmental information according to claim 1 is characterized in that: upgrading right computing formula of keeping away barrier segmental arc curvature in the described step (7) is:

$S_r=\frac{1}{R_r}=\frac{2(y_r-d)}{{\left(x_i\right)}^2+{\left(y_r\right)}^2-d^2}$

Work as y in the formula _rY is worked as in-d 〉=0 o'clock _r-d＜0 o'clock x _i=x _r, a is a width for mobile robot's length, b, $d=\sqrt{{(a/2)}^2+{(b/2)}^2},$ R _rFor keeping away the Y coordinate that hinders the segmental arc center of circle in the right side, its absolute value is the right radius of keeping away the barrier segmental arc to the center of circle, and D is for planning coverage, (x _f, x _r, y _l, y _r) be the boundary coordinate of known obstacle.

If adopt variable period to carry out in the local paths planning, the arc length that then also needs the right side to be kept away the barrier segmental arc is upgraded, and the arc length computing formula is:

$L_r=\min [D,\max (L_r,|R_r{\sin }^{-1}\frac{x_f}{R_r-d}\left|\right)].$

12, according to claim 10 or 11 described mobile robot's local paths planning methods, it is characterized in that: the curvature S that is calculating based on the binary environmental information _lOr S _rLevel off to zero the time, the radius infinity can appear in calculating, and then the barrier segmental arc is kept away and right computing formula of keeping away barrier segmental arc arc length changes in a left side:

L＝min[D，max(L，x _f)]。

Images