CN101231669A - Method for searching optimum route of maze - Google Patents

Abstract

The invention discloses a method which can search the optimal path of a maze, and belongs to the artificial intelligence field. Aiming at the random and complicated maze, the method can automatically search the optimal path and carry out real time corrections to the motion of a robot in the maze. The method comprises a lower machine and an upper machine. The upper machine comprises a DSP as well as a CCD camera and a wireless transmission module which are connected with the DSP, the CCD camera collects the global image of the maze and sends to the DSP after being decoded by a video decoder, and the image is sent to the lower machine through the wireless transmission module after being processed by the DSP. The lower machine comprises a trolley and a control system arranged in the trolley, the bottom of the trolley is connected with wheels and a stepping motor, and the control system comprises a singlechip wireless receive module. The wireless receive module receives the information sent out by the wireless transmission module in the upper machine, and the trolley is driven to move by the stepping motor which is driven by the driving circuit of the stepping motor after the information is analyzed. The invention realizes automatically searching the optimal path of the maze and the real time corrections to the motion of the robot can be carried out.

Description

A kind of method of searching optimum route of maze

Technical field

The present invention is a kind of method of searching optimum route of maze, can carry out from the main search optimal path and revises the motion of robot in the labyrinth in real time complicated labyrinth at random, belongs to artificial intelligence field.

Background technology

The experiment of maze robot system can embody stability, controllability and system's antijamming capability of the key concept of many control theories such as system and path optimal programming or the like intuitively, therefore has a lot of research institutions and university that it is studied.

Present most maze robot all is to carry several photoelectric sensors the white guide line on ground, labyrinth is discerned the labyrinth of walking, walk maze robot (referring to " network technology epoch " 2004 03 phase) as people such as the Qian Zhenyan of Shanghai university of communications design, carried 8 reflective photoelectric sensors ground, labyrinth white line has been detected actions such as the turning that realizes advancing, carried out the labyrinth search according to lefft-hand rule or right-hand rule (being that robot changes all the time to the left or to the right along the labyrinth wall).Finish under the prerequisite of set function, because it is too much to carry sensor, increased manufacturing cost, also make simultaneously the sensor misoperation cause the unsettled probability of system to increase, and this robotlike's lack of wisdom and independence, for the selection shortage cognitive ability of optimal path, the full journey of at every turn all will having walked in the labyrinth just can search out the path, can't solve for walking problem in reconfigurable labyrinth, therefore significant limitation is arranged in actual applications.

A lot of problems just because of its existence all are difficult for solving, miniaturization issues as robot, the planning of optimal path reaches the real-time route correction problem to maze robot, therefore these all become the emphasis and the difficult point of research, people make it can be applied in more wide field at the restricting relation of constantly seeking between them in the hope of finding a kind of many-sided balance.

Summary of the invention

The invention provides a kind of method of searching optimum route of maze, when this robot system is walked in the labyrinth, neither need to use the white line channeling conduct, also do not need to allow robot in the labyrinth, walk and just can search out the path after one time, so the present invention has very big independence and intelligent.Also have the little advantage of volume simultaneously, can in the complicated labyrinth of size as the chessboard, carry out work,

General thought of the present invention: adopt the DSP development board as host computer, use overall camera CCD that the labyrinth image is gathered, with the picture signal that collects deliver to carry out image recognition and path planning (adopting the A* algorithm) in the DSP development board after, produce one group of steering order, wireless sending module and miniature mobile robot by DSP development board end communicate, when robot car body electric control part branch receives the control information that DSP development board wireless sending module sends, the microcontroller of robot is resolved information, what parse is the motion of stepper motor configuration, and microcontroller is walked out the labyrinth smoothly by I/O mouth control step motor drive module and then control dolly.

Concrete technical scheme is as follows:

A kind of minisize maze robot includes host computer and is used for 16 slave computers 15 of walking in the labyrinth.Wherein, host computer includes DSP development board 18 and the CCD camera 17 that is connected with DSP development board 18, wireless sending module 20.CCD camera 17 gathers the global image in whole labyrinth 16 and by sending DSP after the video decoder decodes to, DSP handles the back to image and sends to slave computer 15 by wireless sending module.Described slave computer 15 includes dolly and is arranged on the interior control system of dolly, the bottom of dolly is connected with wheel 11 and stepper motor 10, control system includes single-chip microcomputer and the wireless receiving module 5 that is connected with single-chip microcomputer, stepper motor driving circuit 3, wireless receiving module 5 receives the information that the wireless sending module 20 in the host computer sends, after it is resolved, drive carriage walkings by stepper motor driving circuit 3 drive stepping motor 10.The outlet in the wall in described labyrinth 16, the road in labyrinth, labyrinth and the headstock of dolly, the tailstock are with different color marks.

Described host computer also includes the touch-screen that is connected with DSP, and DSP delivers to screen displaying with the image that CCD camera 17 is gathered.

A kind of method of searching optimum route of maze, this method realizes according to the following steps:

1) CCD camera 17 is gathered all images information in labyrinth 16, and the image information that collects is sent to the CoreA of DSP after by video decoder decodes, the picture format of this moment is 720 * 576YUV422, CoreA gets wherein one from two field signals, the image information of 720 * 288YUV422 just, sending CoreB to after being converted into the image of 320 * 240YUV444, also is that 320 * 240 RGB image is delivered to screen displaying simultaneously with the image transitions of 320 * 240YUV444;

2) CoreB carries out threshold segmentation to the pretreated picture of CoreA, finds out position, labyrinth state and the exit position, labyrinth at dolly place from the colouring information of image, takes out 6 * 8 standard labyrinth on the basis of threshold segmentation;

3) CoreB searches out the optimal path of carriage walking by the A* algorithm in the artificial intelligence field, and each flex point coordinate of optimum route of maze demarcated is a series of impact point;

4) determine the position of dolly in labyrinth 16 by the headstock tailstock of dolly and the different colours information in labyrinth, and generate a cover steering order, this steering order sends to control system in the slave computer 15 by wireless sending module 20;

5) single-chip microcomputer in the slave computer 15 is resolved back control step motor 10 to control signal, and then the control dolly is made corresponding sports in the labyrinth;

Outlet is that destination county will no longer send instruction if dolly has arrived the labyrinth, otherwise repeating step 1～step 5 is walked out the labyrinth up to dolly.

Robot among the present invention does not need to use the white line channeling conduct when walking in the labyrinth, can seek out optimal path and can carry out path modification in real time in the restructural labyrinth.Because the circuit structure that the present invention adopts makes that the robot volume is less, can in the complicated labyrinth of size as the chessboard, walk.

Description of drawings

Fig. 1 is an electromechanical structure design diagram of the present utility model

Fig. 2 is a car body part front view of the present utility model

Fig. 3 is a car body part left view of the present utility model

Fig. 4 is a total system design sketch of the present utility model

Fig. 5 is a power supply design frame chart of the present utility model

Fig. 6 is a system hardware The general frame of the present utility model

Fig. 7 is a car body control section schematic diagram of the present utility model

Fig. 8 is a motor-driven schematic diagram of the present utility model

Fig. 9 is a system of the present utility model general flow chart

Figure 10 is DSP image recognition of the present utility model and route searching sub-process figure

Figure 11 is car body control algolithm sub-process figure of the present utility model

Among the figure: 1, base, 2, pulley base, 3, the step motor drive module, 4, slave computer microcontroller circuit plate, 5, wireless receiving module, 6, the DC-DC module, 7, the serial ports level transferring chip, 8, lithium battery, 9, fuselage, 10, decelerating step motor, 11, two-wheel, 12, the rubber wheel cover tire, 13, axle sleeve, 14, bearing, 15, slave computer, 16, the labyrinth, 17, CCD camera, 18, DSP, 20, wireless sending module.

Embodiment

1～Fig. 8 further specifies specific embodiments of the invention below in conjunction with accompanying drawing.

As shown in Figure 4, present embodiment mainly includes host computer 18 and slave computer 15 two parts, and host computer includes DSP development board 18, CCD camera 17, touch-screen and wireless sending module 20.Slave computer 15 includes car body part and control section, is described respectively below.

Slave computer includes dolly and is arranged on the interior control system of dolly.The structure of dolly such as Fig. 1～shown in Figure 3, include base 1, fuselage 9 and two-wheel 11.

As shown in Figure 1, base 1 is the box structure of interior sky, and the right and left is respectively opened a hole, and is by bearing 14 and axle sleeve 13 that wheel 11 and stepper motor 10 is fixing.Stepper motor 10 is an air-conditioning special retarding stepper motor, links to each other with motor drive ic 3.Two-wheel 11 specifications are identical, are metal wheel hub, and rubber wheel cover tire 12 can increase wheel friction force, help strengthening its ground ability of taking off in smooth road.

Fuselage 9 is the box structure of interior sky, and its control section all is built in the box body.The PVC material that the fuselage working strength is higher alleviates body quality and can better insulate simultaneously.

Control section in the slave computer includes single-chip microcomputer and the wireless receiving module 5, the stepper motor driving circuit that are connected with single-chip microcomputer.What single-chip microcomputer in the present embodiment adopted is the ATmaga8L chip of atmel corp, and this family chip adopts reduced instruction set computer, has at a high speed low cost, advantage such as low-power consumption and being widely used.Single-chip microcomputer and the annexation of motor-drive circuit see Fig. 7, Fig. 8 (about drive identical provide one) for details, the PortC port is connected to step motor drive module: PC3-2 and controls left motor positive and inverse and rotating speed, PC3 is the low level counter-rotating, high level just changes, and PC2 pulse foot control system umber of pulse is controlled left motor rotational angle; PC1-0 controls right motor positive and inverse and rotating speed, and the same PC1 low level counter-rotating is just being changeed during the PC1 high level, and PC0 pulse foot control system umber of pulse is controlled right motor rotational angle.The walking of robot is led by these two driving wheels and two pulley bases and is realized, realizes the turning of robot by the speed discrepancy of driving wheel.Control straight line precision can reach 1.8mm (be each pulse robot advance 1.8mm), and corner accuracy reaches 1.8 degree (but being that 1.8 degree are rotated in each pulse robot original place).Whether car body control section sub-process figure sees Figure 11, at first carries out self check and initialization action when robot powers on, judge to interrupt then arriving, if will resolve the control information that host computer sends and produce steering order control step motor corresponding actions.

The high power capacity 3.7V lithium battery that the power supply of maze robot is adopted, at first voltage is transformed into 12V from 3.7V with two DC-DC power transformation modules, thereby be respectively two stepper motor power supplies, again one road 12V electricity being converted to standard 5V through the three-terminal voltage-stabilizing chip in addition is single-chip microcomputer and wireless module power supply.

Host computer includes DSP development board and CCD camera 17 that is connected with DSP and wireless sending module 20.

Host computer DSP control system control robot walk labyrinth 16 be by overall camera CCD gather the black and white colouring information in labyrinth and in real time detection machine people's different end to end colour codes (blue red circular colour code information) identify, black block is the wall in labyrinth, white blocks is the road in labyrinth, blueness is designated as headstock, redness is designated as the tailstock, green is designated as the labyrinth outlet, can be referring to Fig. 4.Its workflow can be referring to Figure 10, CCD camera 17 at first captured the image information of D1 form when program began, image information is delivered to the CoreA of BF561-DSP development board through ADV7181VideoDecoder, the picture format of this moment is 720 * 576YUV422 form, CoreA gets wherein one from two field signals, the image information of 720 * 288YUV422 just, the image that is converted into 320 * 240YUV444 is further discerned and computing for CoreB, and with the image transitions of 320 * 240YUV444 is 320 * 240 RGB image, for showing on touch-screen.

The CoreA of development board is responsible for man-machine interaction and the image of CCD camera collection is carried out pre-service discern and computing for CoreB, the last operation of CoreA μ cos I I operating system, write peripheral hardware driving and GUI and application program, realized man-machine interaction by a touch-screen.CoreB is responsible for the image information that the CoreA pre-service is good and discerns, mainly be that pretreated picture is carried out threshold segmentation, finding out the dolly position from the colouring information of image (asks for average by the probability filter method to the circular colour code of blueness and is labeled as the headstock coordinate points, red circular mark is asked for average and is labeled as tailstock coordinate points), labyrinth state and exit position, labyrinth, (just the image with 320 * 240YUV444 carries out the image block that the threshold segmentation aftertreatment becomes 6 * 8 40 * 40 size to take out 6 * 8 on the basis of threshold segmentation, and the black after the threshold segmentation is decided to be the wall in labyrinth, white is decided to be the road in labyrinth) the standard labyrinth, just finished the conversion and the work such as search the headstock tailstock and impact point of labyrinth map this moment; At this moment the whole routing informations in the labyrinth have been obtained, this moment, CoreB searched out optimal path by the A* algorithm in the artificial intelligence field (impact point is a major parameter), and the demarcation of each flex point on the optimum route of maze (P1-P6 point and entrance, exit point) coordinate is a series of impact point.

On real-time control algolithm, set up the wide area coordinate system to the labyrinth, set up local coordinate system to car body, (headstock tailstock line is as ordinate at first to calculate the car body local coordinate system, blue circular colour code is a headstock, red circular colour code is the tailstock) with the declinate  of labyrinth wide area coordinate system, calculate car body center and the adjacent target point line angle η in the wide area coordinate system of labyrinth again, and obtain the declinate θ of  and η, wire length with car body center and adjacent target point, thereby generate a cover steering order, this steering order sends to single-chip microcomputer in the slave computer by wireless sending module, single-chip microcomputer is resolved back control step motor to control signal, and then the motion of control fuselage.Outlet is that destination county will no longer execute instruction if robot has arrived the labyrinth, otherwise, repeat above flow process, correctly walk out the labyrinth up to robot.

The communication flow of the BF561 double-core course of work and host computer and slave computer can be referring to Fig. 6, and the total system flow process is referring to Fig. 9.

Circuit structure and physical construction that the present invention adopts, make that the robot volume is small and exquisite, and satisfied the accuracy requirement of walking big or small labyrinth as the chessboard, and under the support of software programming, adopt the advanced algorithm in the artificial intelligence field that maze path is carried out Real time identification and correction.

Characteristics of the present utility model are the microminiature of maze robot, and for maze robot has designed new structure, be divided into mechanical part and control section two-layer up and down, not only reduce volume but also improved overall precision and stability, simultaneously, control system has also embodied robustness and the antijamming capability of maze robot aspect software, the path planning and the real-time route correction problem in restructural labyrinth have been solved, it can be applied to the research and the experiment of robotics and control theory subject as novel research object, and can be widely used in intelligent transportation, microminiature multi-purpose robot's research.

The maze robot that a kind of size that provides in the present embodiment is small based on DSP control, can be autonomous walk out specific zone, labyrinth, can replace the combatant to carry out Special Warfare Mission or execute the task in the place of artificial operational difficulty, its core technology also can be widely used in intelligent transportation field, each vehicle can be finished automatic driving under the commander of vision system, the whole path planning of vision system will improve the traffic congestion situation greatly, and reduce the generation of traffic hazard; Be applied to military field, utilize GPS to locate it and can finish military target identification location, search automatically, but and the remote independent control robot disturb enemy army etc.; Be applied to the Urban Search and Rescue field, it can be sought out best rescue route and arrive the scene of the accident rapidly in complex-terrain, and counterweight injury personnel carry out preliminary emergency processing, carries out rescue work thereby help to search and rescue the team member to greatest extent; Because it has a extensive future, also have certain positive effect for the scientific research of China, as microminiature multi-purpose robot's research.

Claims (1)

1. the method for a searching optimum route of maze is characterized in that, this method realizes according to the following steps:

1) CCD camera (17) is gathered all images information in labyrinth (16), and the image information that collects is sent to the CoreA of DSP after by video decoder decodes, the picture format of this moment is 720 * 576YUV422, CoreA gets wherein one from two field signals, the image information of 720 * 288YUV422 just, sending CoreB to after being converted into the image of 320 * 240YUV444, also is that 320 * 240 RGB image is delivered to screen displaying simultaneously with the image transitions of 320 * 240YUV444;

The outlet in the wall in described labyrinth (16), the road in labyrinth, labyrinth and the headstock of dolly, the tailstock are with different color marks;

2) CoreB carries out threshold segmentation to the pretreated picture of CoreA, finds out position, labyrinth state and the exit position, labyrinth at dolly place from the colouring information of image, takes out 6 * 8 standard labyrinth on the basis of threshold segmentation;

3) CoreB searches out the optimal path of carriage walking by the A* algorithm in the artificial intelligence field, and each flex point coordinate of optimum route of maze demarcated is a series of impact point;

4) determine the position of dolly in labyrinth (16) by the headstock tailstock of dolly and the different colours information in labyrinth, and generate a cover steering order, this steering order sends to control system in the slave computer (15) by wireless sending module (20);

5) single-chip microcomputer in the slave computer (15) is resolved back control step motor (10) to control signal, and then the control dolly is made corresponding sports in the labyrinth;

Outlet is that destination county will no longer send instruction if dolly has arrived the labyrinth, otherwise repeating step 1～step 5 is walked out the labyrinth up to dolly.

Images