CN110109456A - A kind of trolley automatic Pilot method - Google Patents
A kind of trolley automatic Pilot method Download PDFInfo
- Publication number
- CN110109456A CN110109456A CN201910339039.9A CN201910339039A CN110109456A CN 110109456 A CN110109456 A CN 110109456A CN 201910339039 A CN201910339039 A CN 201910339039A CN 110109456 A CN110109456 A CN 110109456A
- Authority
- CN
- China
- Prior art keywords
- motor
- automatic pilot
- motors
- drive circuit
- image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 24
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 19
- 238000012549 training Methods 0.000 claims abstract description 9
- 238000013135 deep learning Methods 0.000 claims abstract description 7
- 235000011034 Rubus glaucus Nutrition 0.000 claims description 28
- 235000009122 Rubus idaeus Nutrition 0.000 claims description 28
- 240000007651 Rubus glaucus Species 0.000 claims description 25
- 239000003990 capacitor Substances 0.000 claims description 18
- 230000001360 synchronised effect Effects 0.000 claims description 10
- 230000005611 electricity Effects 0.000 claims description 7
- HBBGRARXTFLTSG-UHFFFAOYSA-N Lithium ion Chemical compound [Li+] HBBGRARXTFLTSG-UHFFFAOYSA-N 0.000 claims description 3
- 229910001416 lithium ion Inorganic materials 0.000 claims description 3
- 238000004891 communication Methods 0.000 claims description 2
- 230000001537 neural effect Effects 0.000 claims 1
- 238000005516 engineering process Methods 0.000 abstract description 3
- 238000013480 data collection Methods 0.000 abstract description 2
- 230000007547 defect Effects 0.000 abstract description 2
- 238000010586 diagram Methods 0.000 description 4
- 244000235659 Rubus idaeus Species 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 238000009395 breeding Methods 0.000 description 1
- 230000001488 breeding effect Effects 0.000 description 1
- 239000003054 catalyst Substances 0.000 description 1
- 230000005484 gravity Effects 0.000 description 1
- 210000002364 input neuron Anatomy 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 238000003475 lamination Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 210000003205 muscle Anatomy 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 238000003825 pressing Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
- G05D1/00—Control of position, course, altitude or attitude of land, water, air or space vehicles, e.g. using automatic pilots
- G05D1/02—Control of position or course in two dimensions
- G05D1/021—Control of position or course in two dimensions specially adapted to land vehicles
- G05D1/0231—Control of position or course in two dimensions specially adapted to land vehicles using optical position detecting means
- G05D1/0246—Control of position or course in two dimensions specially adapted to land vehicles using optical position detecting means using a video camera in combination with image processing means
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Multimedia (AREA)
- Electromagnetism (AREA)
- Aviation & Aerospace Engineering (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- General Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention discloses a kind of trolley automatic Pilot methods.Automatic Pilot technology is exactly the combination of deep learning and car steering, this is the project for being researched and developed and being possessed great potential in the world.The automatic Pilot vehicle that the present invention uses includes vehicle body, wheel, motor and control module.Step of the invention is as follows: one, building convolutional neural networks.Two, automatic Pilot vehicle is placed on work-yard, and drives automatic Pilot vehicle to travel along target route by way of artificially controlling, to establish path profile image set.Three, with the convolutional neural networks established in path image collection training step 1.Four, automatic Pilot vehicle is placed on the destination path of work-yard and is travelled.The present invention can store multiple path data collection simultaneously so that trolley can in a plurality of track free switching, overcome it is conventional can only be along the defect of fixation locus along track automatic Pilot trolley.
Description
Technical field
The invention belongs to technical field of automatic control, and in particular to a kind of trolley automatic Pilot method.
Background technique
As deep learning becomes new optimal algorithm in computer vision field in 2012-2013, and finally exist
All become optimal algorithm on all perception tasks, industry starts center of gravity being biased to it, and artificial intelligence belonging to deep learning because
For the leap of hardware, three kinds of algorithm, data technologies, the preferably period developed in decades has been come into.Utilize depth
Practise algorithm can agricultural breeding, security protection insurance, intelligence manufacture, car steering and in terms of have deep application, not only
Efficiency can be improved and solve existing problem, can more expand out new industry.Wherein automatic Pilot technology is exactly deep learning and vapour
The combination that vehicle drives, this is the project for being researched and developed and being possessed great potential in the world.All sectors of society is for this skill
Art all entertains very big concern and expectation, is mainly reflected in the following aspects:
1. individual demand: driving, which is substantially one, needs that permanent visual analysis and the mechanical of muscle response is kept to repeat to transport
It is dynamic, this and inhuman strong point is also easy to make one to be fed up with, and this exactly machine is good at.
2. the market demand: automatic Pilot can be applied not only to traffic trip, all will for logistics transportation, urban planning etc.
Generate the innovation of essence.
3. capital requirement: having there is a large amount of capital to flow into big data, novel sensor, depth that automatic Pilot is related in front
Learning art etc., the flowing of capital are often the catalyst of industry development most critical, have investment just to may require that return, investor
It can try one's best and push the development of automatic Pilot.
4. social demand: from the point of view of the country such as the U.S. at this stage, Germany, China is directed to the policy of automatic Pilot, society is needed
Ask strong, national attention degree is high.
Summary of the invention
The purpose of the present invention is to provide a kind of methods for realizing motor-driven carrier automatic Pilot.
A kind of trolley automatic Pilot method of the present invention, the automatic Pilot vehicle of use include vehicle body, wheel, motor and control mould
Block.Four wheels are in pairs on the vehicle body of centering bearing.Four wheels are respectively driven by four motors.The installation of CSI camera
In the head end of vehicle body.Control module includes raspberry pie system, CSI camera and motor-drive circuit.The shooting of CSI camera is shone
Piece is simultaneously transferred to raspberry pie system.Motor-drive circuit driving motor under the control of raspberry pie system by motor driver.
The trolley automatic Pilot method is specific as follows:
Step 1: building convolutional neural networks.The convolutional neural networks include the input layer being arranged successively, five volumes
Lamination, two ill-mannered state layer and two dense layers.
Step 2: automatic Pilot vehicle is placed on work-yard, and automatic Pilot is driven by way of artificially controlling
Vehicle is travelled along target route.Camera every preset time period shoots a path image, group in automatic Pilot vehicle driving process
At path profile image set.And each path image is divided into four groups, it is respectively corresponded with four labels.Label be respectively advance, turn left,
It turns right, stop.
If the synchronous constant speed of four motors rotates forward path image when shooting, it is corresponding which is included into advance label
Image group；If path image is located at the revolving speed of two motors on the left of vehicle body lower than two motors on the right side of vehicle body when shooting
Revolving speed, then the path image is included into the corresponding image group of left-hand rotation label；If path image is located on the right side of vehicle body when shooting
The revolving speed of two motors is lower than the revolving speed of two motors on the left of vehicle body, then the path image is included into the corresponding figure of right-hand rotation label
As group；If four motors stall path image when shooting, which, which is included into, stops the corresponding image group of label.
Step 3: with the convolutional neural networks established in path image collection training step 1.
Step 4: automatic Pilot vehicle is placed on the destination path of work-yard；Camera persistently shoots tested photo.
Tested photo is inputted into convolutional neural networks；Convolutional neural networks are identified and are classified to by altimetric image, are determined by altimetric image
With advance, left-hand rotation, right-hand rotation, the stopping respective matching probability of label.
If the synchronous constant speed of four motors rotates forward by the matching probability highest of altimetric image and advance label.If by altimetric image
With the matching probability highest of left-hand rotation label, then four motors are synchronous rotates forward, and the revolving speed of two motors on the left of vehicle body is lower than vehicle
The revolving speed of two motors on the right side of body；If four motors synchronize just by the matching probability highest of altimetric image and right-hand rotation label
Turn, and the revolving speed of two motors on the right side of vehicle body is lower than the revolving speed of two motors on the left of vehicle body.If by altimetric image and stopping marking
The matching probability highest of label, then four motors stall, and are considered as have arrived at terminal at this time.
Further, the convolutional neural networks in step 1 are built by keras deep learning library.
Further, the input layer uses lambda layers.First three convolutional layer is 5 × 5 convolutional layers；Latter two convolution
Layer is 3 × 3 convolutional layers.The convolution kernel of five convolutional layer final outputs is 3 × 3 matrixes of 64 dimensions.First ill-mannered state layer uses
Dropout layers.Second ill-mannered state layer uses flatten layers.Two dense layers are full articulamentum.
Further, the target route is in a ring.
Further, in step 2, the resolution ratio of all path images is 120 × 160 in path profile image set.
Further, the motor-drive circuit includes motor driver.The model of motor driver
TB6612FNG.One end of the equal connecting resistance R1 of 3,4,9 and 10 pins of motor driver.The other end of resistance R1 is grounded.Motor
13,14,24 pins of driver connect one end of capacitor C1, capacitor C2 and capacitor C3.Capacitor C1, capacitor C2 and capacitor C3's
The other end connects external voltage.15 and 23 pins of motor driver link together, the revolving speed control as motor-drive circuit
End processed；17 and 21 pins of motor driver link together, the first course changing control end as motor-drive circuit；Motor
16 and 22 pins of driver link together, the second course changing control end as motor-drive circuit；Motor driver 1,
First motor control interface of 5 pins as motor-drive circuit.8,12 pins of motor driver are as motor-drive circuit
The second motor control interface.
There are two the motor-drive circuit is total.The first motor control interface of one of motor-drive circuit,
The interface of two motor control interfaces and two motors for being located at the same side is separately connected.The first of another motor-drive circuit
The interface of motor control interface, the second motor control interface and two motors positioned at the other side is separately connected.
Further, the model BCM2837 of the raspberry pie system.Raspberry pie system 6,9,14,20,25,30,
34 and 39 pins are grounded, and 2 and 4 pins connect external voltage.First I/O mouthfuls, the 2nd I/O mouthfuls, the 3rd I/ of raspberry pie system
O mouthfuls are separately connected with the revolving speed control terminal of one of motor-drive circuit, the first course changing control end, the second course changing control end.
The 4th I/O mouthfuls of raspberry pie system, the 5th I/O mouthfuls, the 6th I/O mouthfuls with the revolving speed control terminal of another motor-drive circuit,
One course changing control end, the second course changing control end are separately connected.
Further, the automatic Pilot vehicle further includes battery.The battery uses lithium ion battery, for control
Module and motor power supply.
Further, 15 terminals of 1 to 15 pin of the CSI camera and CSI interface are separately connected.CSI connects
Mouth is plugged on the utilizing camera interface of raspberry pie system.
Further, the raspberry pie system and host computer pass through wireless communication.
The invention has the advantages that:
1, the present invention can be not laid with physics track item in the case where, realize trolley it is automatic along track advance, in turn
The track laying cost of the application scenarios such as AGV trolley self-navigation can be substantially reduced.
2, the present invention can store multiple path data collection simultaneously, so that trolley can be in a plurality of track certainly
By switching, overcome it is conventional can only be along the defect of fixation locus along track automatic Pilot trolley.
Detailed description of the invention
Fig. 1 is system block diagram of the invention；
Fig. 2 is the circuit diagram of motor-drive circuit in the present invention；
Fig. 3 is the circuit connection diagram of raspberry pie system in the present invention；
Fig. 4 is the system block diagram of convolutional neural networks in the present invention.
Specific embodiment
Below in conjunction with attached drawing, the invention will be further described.
As shown in Figure 1, a kind of trolley automatic Pilot method, the automatic Pilot vehicle of use includes that vehicle body, wheel, electricity are mechanical, electrical
Pond 2 and control module.Four wheels are in pairs on the vehicle body of centering bearing.Four wheels are respectively driven by four motors.Electricity
Pond 2 uses lithium ion battery, is that control module and motor are powered.CSI camera 5 is fixed on the head end of vehicle body.Control module
Including raspberry pie system 3, CSI camera 5 and motor-drive circuit 4.Raspberry pie system 3 and host computer 1 are logical by wireless network
Letter.CSI camera 5 shoots photo and is transferred to raspberry pie system 3.Motor-drive circuit 4 is by motor driver U1 in raspberry
Factions' system 3 controls lower driving motor.
As shown in Fig. 2, motor-drive circuit 4 includes motor driver U1.The model of motor driver U1
TB6612FNG.One end of the equal connecting resistance R1 of 3,4,9 and 10 pins of motor driver U1.The other end of resistance R1 is grounded.Electricity
13,14,24 pins of machine driver U1 connect one end of capacitor C1, capacitor C2 and capacitor C3.Capacitor C1, capacitor C2 and capacitor
The other end of C3 connects external voltage VCC (5V voltage).15 and 23 pins of motor driver U1 link together, as electricity
The revolving speed control terminal of drive circuit 4；17 and 21 pins of motor driver U1 link together, as motor-drive circuit 4
The first course changing control end；16 and 22 pins of motor driver U1 link together, second as motor-drive circuit 4
Course changing control end；First motor control interface of 1,5 pins of motor driver U1 as motor-drive circuit 4.Motor driven
Second motor control interface of 8,12 pins of device U1 as motor-drive circuit 4.
There are two motor-drive circuit 4 is total.The first motor control interface of one of motor-drive circuit 4, the second electricity
Machine control interface and the interface for two motors for being located at the same side are separately connected.The first motor of another motor-drive circuit 4
The interface of control interface, the second motor control interface and two motors positioned at the other side is separately connected.Same motor driven electricity
Two motors that road 4 connects are denoted as M1, M2 in Fig. 2.The revolving speed control terminal of one of motor-drive circuit 4,
One course changing control end, the second course changing control end are denoted as PWM, IN1, IN2.The revolving speed of another motor-drive circuit 4 controls
End, the first course changing control end, the second course changing control end are denoted as PWM', IN1', IN2'.
As shown in figure 3, the model BCM2837 of raspberry pie system.Raspberry pie system 6,9,14,20,25,30,34 and
39 pins are grounded, and 2 and 4 pins connect external voltage.The first I/O mouthfuls (36 pin) of raspberry pie system, the 2nd I/O mouthfuls (38
Pin), the 3rd I/O mouthfuls (40 pin) with the revolving speed control terminal PWM of one of motor-drive circuit 4, the first course changing control end
IN1, the second course changing control end IN2 are separately connected.The 4th I/O mouthfuls (33 pin) of raspberry pie system, the 5th I/O mouthfuls (35 draw
Foot), the 6th I/O mouthfuls (37 pin) with the revolving speed control terminal PWM' of another motor-drive circuit 4, the first course changing control end
IN1', the second course changing control end IN2' are separately connected.15 of 1 to 15 pin of CSI camera 5 and CSI interface (plug connector)
Terminals are separately connected.CSI interface is plugged on the utilizing camera interface of raspberry pie system 3.
The trolley automatic Pilot method is specific as follows:
Step 1: convolutional neural networks are built in keras deep learning library.
Neural network of the invention is a kind of simplification algorithm of structure based on sequence, this algorithm is applied to trolley, can be with
There are good stability and robustness.Algorithm structure figure is as shown in the figure.
As shown in figure 4, convolutional neural networks include the input layer being arranged successively, five convolutional layers, two ill-mannered state layer and
Two dense layers.Input layer uses lambda layers.In lambda layers, the shape of each element of color image data tensor is
(120,160,3), and value interval, in [0,255], biggish gradient updating, causes network that can not restrain, to every in order to prevent
A input element, which is normalized, makes its range in [- 0.5,0.5], so that it is adapted to the network structure of same type, and add
Training of the fast model on GPU.
First three convolutional layer is 5 × 5 convolutional layers (its convolution window is 2 × 2)；Latter two convolutional layer is 3 × 3 convolutional layers.
The convolution kernel of five convolutional layer final outputs is 3 × 3 matrixes of 64 dimensions.First ill-mannered state layer uses dropout layers,
Disconnected at random when each undated parameter in training process 50% input neuron connection, prevent model start study only and
The relevant mode of training data, this model are wrong for verify data.Second ill-mannered state layer uses
Flatten layers, it is used to input " pressing ", that is, one-dimensional data is finally entered, used in from convolutional layer to the mistake of full articulamentum
It crosses.Two dense layers are full articulamentum, carry out elu (dot (input, kernel)+bias) operation to tensor, i.e., will input
Dot product is carried out with the weight tensor of this layer, then adds bias vector, this two layers feature for being used to extract convolutional layer carries out
Training and final output steering controling signal.Elu is activation primitive, it obtains the output of this layer and send as input
To next layer.Since simple linear operation keeps each layer of hypothesis space very limited, multiple linear layers, which stack, to be realized still
It is linear operation, introduces nonlinear characteristic with activation primitive, so that it may is abundant to assume that space makes full use of multilayer to indicate excellent
Gesture.Second dense layer can export the tensor of 4 elements, and the probability for indicating that picture is assigned under this 4 labels is more respectively
It is few, and probability summation is 1.
Step 2: automatic Pilot vehicle is placed on work-yard, and automatic Pilot is driven by way of artificially controlling
Vehicle is multiple along target route traveling.Target route is in a ring (i.e. beginning and end is the same point).Automatic Pilot garage crosses
Camera shoots a path image every T time in journey.The resolution ratio of all path images is adjusted to 120 × 160,
And form path profile image set.And each path image is divided into four groups, it is respectively corresponded with four labels.Label is respectively to advance, is left
Turn, turn right, stop.
If the synchronous constant speed of four motors rotates forward path image when shooting, it is corresponding which is included into advance label
Image group；If path image is located at the revolving speed of two motors on the left of vehicle body lower than two motors on the right side of vehicle body when shooting
Revolving speed, then the path image is included into the corresponding image group of left-hand rotation label；If path image is located on the right side of vehicle body when shooting
The revolving speed of two motors is lower than the revolving speed of two motors on the left of vehicle body, then the path image is included into the corresponding figure of right-hand rotation label
As group；If four motors stall path image when shooting, which, which is included into, stops the corresponding image group of label
(image is target route).
And then path image and driving direction is made to form corresponding relationship by way of setting label, every a pair of parameter is all
Example required for training learning model.
Step 3: with path image collection training convolutional neural networks.
Step 4: automatic Pilot vehicle is placed on the destination path of work-yard；Camera persistently shoots tested photo.
Tested photo is inputted into convolutional neural networks；Convolutional neural networks are identified and are classified to by altimetric image, are determined by altimetric image
With advance, left-hand rotation, right-hand rotation, the stopping respective matching probability of label.
If the synchronous constant speed of four motors rotates forward by the matching probability highest of altimetric image and advance label.If by altimetric image
With the matching probability highest of left-hand rotation label, then four motors are synchronous rotates forward, and the revolving speed of two motors on the left of vehicle body is lower than vehicle
The revolving speed of two motors on the right side of body；If four motors synchronize just by the matching probability highest of altimetric image and right-hand rotation label
Turn, and the revolving speed of two motors on the right side of vehicle body is lower than the revolving speed of two motors on the left of vehicle body.If by altimetric image and stopping marking
The matching probability highest of label, then four motors stall, and are considered as have arrived at terminal at this time.
Claims (10)
1. a kind of trolley automatic Pilot method, the automatic Pilot vehicle of use includes vehicle body, wheel, motor and control module；It is special
Sign is: four wheels are in pairs on the vehicle body of centering bearing；Four wheels are respectively driven by four motors；CSI camera
It is mounted on the head end of vehicle body；Control module includes raspberry pie system, CSI camera and motor-drive circuit；The shooting of CSI camera
Photo is simultaneously transferred to raspberry pie system；Motor-drive circuit driving motor under the control of raspberry pie system by motor driver；
The trolley automatic Pilot method is specific as follows:
Step 1: building convolutional neural networks；The convolutional neural networks include the input layer being arranged successively, five convolution
Layer, two ill-mannered state layer and two dense layers；
Step 2: automatic Pilot vehicle is placed on work-yard, and automatic Pilot vehicle edge is driven by way of artificially controlling
Target route traveling；Camera every preset time period shoots a path image in automatic Pilot vehicle driving process, forms road
Diameter image set；And each path image is divided into four groups, it is respectively corresponded with four labels；Label be respectively advance, turn left, turn right,
Stop；
If the synchronous constant speed of four motors rotates forward path image when shooting, which is included into the corresponding image of advance label
Group；If the revolving speed that path image is located at two motors on the left of vehicle body when shooting turns lower than two motors on the right side of vehicle body
Speed, then the path image is included into the corresponding image group of left-hand rotation label；If path image is located at two on the right side of vehicle body when shooting
The revolving speed of motor is lower than the revolving speed of two motors on the left of vehicle body, then the path image is included into the corresponding image group of right-hand rotation label；
If four motors stall path image when shooting, which, which is included into, stops the corresponding image group of label；
Step 3: with the convolutional neural networks established in path image collection training step 1；
Step 4: automatic Pilot vehicle is placed on the destination path of work-yard；Camera persistently shoots tested photo；It will be by
It surveys photo and inputs convolutional neural networks；Convolutional neural networks are identified and are classified to by altimetric image, are determined by altimetric image with before
Into, turn left, turn right, stop the respective matching probability of label；
If the synchronous constant speed of four motors rotates forward by the matching probability highest of altimetric image and advance label；If by altimetric image and a left side
Turn the matching probability highest of label, then four motors are synchronous rotates forward, and the revolving speed of two motors on the left of vehicle body is right lower than vehicle body
The revolving speed of two motors of side；If four motors are synchronous to be rotated forward, and vehicle by the matching probability highest of altimetric image and right-hand rotation label
Revolving speed of the revolving speed of two motors on the right side of body lower than two motors on the left of vehicle body；If by the matching of altimetric image and stopping label
Probability highest, then four motors stall, and are considered as have arrived at terminal at this time.
2. a kind of trolley automatic Pilot method according to claim 1, it is characterised in that: the convolutional Neural net in step 1
Network is built by keras deep learning library.
3. a kind of trolley automatic Pilot method according to claim 1, it is characterised in that: the input layer uses
Lambda layers；First three convolutional layer is 5 × 5 convolutional layers；Latter two convolutional layer is 3 × 3 convolutional layers；Five convolutional layer final outputs
Convolution kernel be 64 dimension 3 × 3 matrixes；First ill-mannered state layer uses dropout layers；Second ill-mannered state layer uses
Flatten layers；Two dense layers are full articulamentum.
4. a kind of trolley automatic Pilot method according to claim 1, it is characterised in that: the target route is in ring
Shape.
5. a kind of trolley automatic Pilot method according to claim 1, it is characterised in that: in step 2, path profile image set
The resolution ratio of interior all path images is 120 × 160.
6. a kind of trolley automatic Pilot method according to claim 1, it is characterised in that: the motor-drive circuit packet
Include motor driver；The model TB6612FNG of motor driver；The equal connecting resistance R1 of 3,4,9 and 10 pins of motor driver
One end；The other end of resistance R1 is grounded；13,14,24 pins of motor driver meet capacitor C1, capacitor C2 and capacitor C3
One end；The other end of capacitor C1, capacitor C2 and capacitor C3 connect external voltage；15 and 23 pins of motor driver are connected to one
It rises, the revolving speed control terminal as motor-drive circuit；17 and 21 pins of motor driver link together, as motor driven
First course changing control end of circuit；16 and 22 pins of motor driver link together, second as motor-drive circuit
Course changing control end；First motor control interface of 1,5 pins of motor driver as motor-drive circuit；Motor driver
8, second motor control interface of 12 pins as motor-drive circuit；
There are two the motor-drive circuit is total；The first motor control interface of one of motor-drive circuit, the second electricity
Machine control interface and the interface for two motors for being located at the same side are separately connected；The first motor control of another motor-drive circuit
Interface processed, the second motor control interface and it is located at the interface of two motors of the other side and is separately connected.
7. a kind of trolley automatic Pilot method according to claim 6, it is characterised in that: the model of the raspberry pie system
For BCM2837；6,9,14,20,25,30,34 and 39 pins of raspberry pie system are grounded, and 2 and 4 pins connect external voltage；
The first I/O mouthfuls of raspberry pie system, the 2nd I/O mouthfuls, the 3rd I/O mouthfuls with the revolving speed control terminal of one of motor-drive circuit,
First course changing control end, the second course changing control end are separately connected；4th I/O mouthfuls, the 5th I/O mouthfuls, the 6th I/O of raspberry pie system
Mouth is separately connected with the revolving speed control terminal of another motor-drive circuit, the first course changing control end, the second course changing control end.
8. a kind of trolley automatic Pilot method according to claim 1, it is characterised in that: the automatic Pilot vehicle also wraps
Include battery；The battery uses lithium ion battery, is that control module and motor are powered.
9. a kind of trolley automatic Pilot method according to claim 1, it is characterised in that: the 1 to 15 of the CSI camera
15 terminals of pin and CSI interface are separately connected；CSI interface is plugged on the utilizing camera interface of raspberry pie system.
10. a kind of trolley automatic Pilot method according to claim 1, it is characterised in that: the raspberry pie system with
Host computer passes through wireless communication.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910339039.9A CN110109456A (en) | 2019-04-25 | 2019-04-25 | A kind of trolley automatic Pilot method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910339039.9A CN110109456A (en) | 2019-04-25 | 2019-04-25 | A kind of trolley automatic Pilot method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110109456A true CN110109456A (en) | 2019-08-09 |
Family
ID=67486688
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910339039.9A Pending CN110109456A (en) | 2019-04-25 | 2019-04-25 | A kind of trolley automatic Pilot method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110109456A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112308922A (en) * | 2020-11-25 | 2021-02-02 | 南京鸿昌智能货架有限公司 | Deep learning-based automatic goods-searching positioning method and system for shuttle |
| WO2023035462A1 (en) * | 2021-09-10 | 2023-03-16 | 中国科学院深圳先进技术研究院 | Pulse neural network-based vehicle wheel speed self-regulating method and self-regulating system |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20180239969A1 (en) * | 2017-02-23 | 2018-08-23 | Ford Global Technologies, Llc | Free Space Detection Using Monocular Camera and Deep Learning |
| CN108664028A (en) * | 2018-05-21 | 2018-10-16 | 南昌航空大学 | Convenient for the omnidirectional vision intelligent carriage of secondary development |
-
2019
- 2019-04-25 CN CN201910339039.9A patent/CN110109456A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20180239969A1 (en) * | 2017-02-23 | 2018-08-23 | Ford Global Technologies, Llc | Free Space Detection Using Monocular Camera and Deep Learning |
| CN108664028A (en) * | 2018-05-21 | 2018-10-16 | 南昌航空大学 | Convenient for the omnidirectional vision intelligent carriage of secondary development |
Non-Patent Citations (4)
| Title |
|---|
| 张亚初 等: "基于双流卷积神经网络的智能小车避障算法研究", 《新技术新工艺》 * |
| 胡志超 等: "基于树莓派的移动监控小车设计", 《科技广场》 * |
| 邵彩幸 等: "基于嵌入式与OpenCV的目标识别系统研究", 《西南民族大学学报(自然科学版)》 * |
| 钟皇平 等: "基于卷积神经网络的自动行驶小车研究与设计", 《杭州电子科技大学学报(自然科学版)》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112308922A (en) * | 2020-11-25 | 2021-02-02 | 南京鸿昌智能货架有限公司 | Deep learning-based automatic goods-searching positioning method and system for shuttle |
| WO2023035462A1 (en) * | 2021-09-10 | 2023-03-16 | 中国科学院深圳先进技术研究院 | Pulse neural network-based vehicle wheel speed self-regulating method and self-regulating system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107571867A (en) | Method and apparatus for controlling automatic driving vehicle | |
| CN106875511A (en) | A kind of method for learning driving style based on own coding regularization network | |
| CN110109456A (en) | A kind of trolley automatic Pilot method | |
| CN202351704U (en) | Tracking intelligent car device based on electromagnetic navigation | |
| CN110085078B (en) | Bench track system for intelligent driving auxiliary experiment and practical training teaching of intelligent network-connected automobile | |
| CN104680914A (en) | Intelligent traffic scheduling real object simulation system based on miniature autonomous unmanned vehicle | |
| CN103996283B (en) | Based on many vehicles two-way lane intelligent transportation analogue system and the method for Zigbee network | |
| CN108177693A (en) | Wheel hub drives the electronic differential control system of electric vehicle | |
| CN109271924A (en) | Image processing method and image processing apparatus | |
| CN103021221A (en) | Simulation system and simulation method for virtual driving behavior of drivers of subway trains | |
| CN115048972B (en) | Traffic scene deconstruction classification method and virtual-real combined automatic driving test method | |
| CN110534009A (en) | A kind of unmanned course teaching aid of artificial intelligence | |
| CN108549370A (en) | Collecting method and harvester | |
| CN103500527A (en) | Micro intelligent car framework for intelligent traffic hardware online simulation system | |
| Duan et al. | Encoding distributional soft actor-critic for autonomous driving in multi-lane scenarios | |
| Gajjar et al. | A comprehensive study on lane detecting autonomous car using computer vision | |
| Gao et al. | Autonomous driving based on modified sac algorithm through imitation learning pretraining | |
| CN202351636U (en) | Double-CCD based automatic parking device of intelligent model car | |
| CN202615678U (en) | Teaching experimental device based on camera intelligent vehicle system | |
| CN204066477U (en) | A kind of teaching test carriage | |
| CN106327964A (en) | Teaching experiment trolley tracking system | |
| Wu et al. | A deep reinforcement learning based car following model for electric vehicle | |
| CN207197560U (en) | A kind of traffic mark identifies test device | |
| CN210091429U (en) | Rack rail system for intelligent driving auxiliary experiment and practical training teaching of intelligent networked automobile | |
| CN206781742U (en) | A kind of intelligent vehicle that can be tracked with avoidance |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190809 |