CN105059213A - Intelligent car following control system and method - Google Patents
Intelligent car following control system and method Download PDFInfo
- Publication number
- CN105059213A CN105059213A CN201510489170.5A CN201510489170A CN105059213A CN 105059213 A CN105059213 A CN 105059213A CN 201510489170 A CN201510489170 A CN 201510489170A CN 105059213 A CN105059213 A CN 105059213A
- Authority
- CN
- China
- Prior art keywords
- neural network
- actual spacing
- car
- cost function
- vehicle
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 37
- 230000002708 enhancing effect Effects 0.000 claims abstract description 53
- 230000008859 change Effects 0.000 claims abstract description 11
- 238000013528 artificial neural network Methods 0.000 claims description 153
- 230000009471 action Effects 0.000 claims description 67
- 238000011156 evaluation Methods 0.000 claims description 32
- 238000005728 strengthening Methods 0.000 claims description 31
- 238000001514 detection method Methods 0.000 claims description 8
- 238000011897 real-time detection Methods 0.000 claims description 2
- 230000001133 acceleration Effects 0.000 abstract 1
- 230000006870 function Effects 0.000 description 50
- 230000008569 process Effects 0.000 description 10
- 230000006978 adaptation Effects 0.000 description 5
- 230000001105 regulatory effect Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 206010039203 Road traffic accident Diseases 0.000 description 2
- 230000006399 behavior Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 230000007274 generation of a signal involved in cell-cell signaling Effects 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 102100029469 WD repeat and HMG-box DNA-binding protein 1 Human genes 0.000 description 1
- 101710097421 WD repeat and HMG-box DNA-binding protein 1 Proteins 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000000903 blocking effect Effects 0.000 description 1
- 230000001276 controlling effect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000012886 linear function Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60R—VEHICLES, VEHICLE FITTINGS, OR VEHICLE PARTS, NOT OTHERWISE PROVIDED FOR
- B60R16/00—Electric or fluid circuits specially adapted for vehicles and not otherwise provided for; Arrangement of elements of electric or fluid circuits specially adapted for vehicles and not otherwise provided for
- B60R16/02—Electric or fluid circuits specially adapted for vehicles and not otherwise provided for; Arrangement of elements of electric or fluid circuits specially adapted for vehicles and not otherwise provided for electric constitutive elements
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60K—ARRANGEMENT OR MOUNTING OF PROPULSION UNITS OR OF TRANSMISSIONS IN VEHICLES; ARRANGEMENT OR MOUNTING OF PLURAL DIVERSE PRIME-MOVERS IN VEHICLES; AUXILIARY DRIVES FOR VEHICLES; INSTRUMENTATION OR DASHBOARDS FOR VEHICLES; ARRANGEMENTS IN CONNECTION WITH COOLING, AIR INTAKE, GAS EXHAUST OR FUEL SUPPLY OF PROPULSION UNITS IN VEHICLES
- B60K31/00—Vehicle fittings, acting on a single sub-unit only, for automatically controlling vehicle speed, i.e. preventing speed from exceeding an arbitrarily established velocity or maintaining speed at a particular velocity, as selected by the vehicle operator
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B60—VEHICLES IN GENERAL
- B60K—ARRANGEMENT OR MOUNTING OF PROPULSION UNITS OR OF TRANSMISSIONS IN VEHICLES; ARRANGEMENT OR MOUNTING OF PLURAL DIVERSE PRIME-MOVERS IN VEHICLES; AUXILIARY DRIVES FOR VEHICLES; INSTRUMENTATION OR DASHBOARDS FOR VEHICLES; ARRANGEMENTS IN CONNECTION WITH COOLING, AIR INTAKE, GAS EXHAUST OR FUEL SUPPLY OF PROPULSION UNITS IN VEHICLES
- B60K2310/00—Arrangements, adaptations or methods for cruise controls
- B60K2310/24—Speed setting methods
Abstract
The invention discloses an intelligent car following control system and method, and belongs to the technical field of car active safety. The intelligent car following control system comprises a detecting module, a learning enhancing module and a car following module. The detecting module is used for detecting the actual car distance between a car and a car ahead in real time and generating an enhancing signal corresponding to the actual car distance according to the actual car distance, and the enhancing signal is used for showing the deviation amplitude between the actual car distance and the set ideal car distance. The learning enhancing module is used for adopting a learning enhancing mode for determining the adjusting amplitude of the running speed of the car according to the actual car distance and the enhancing signal. The car following module is used for adjusting the running speed of the car according to the adjusting amplitude to change the actual car distance. The situation that due to the fact that acceleration and speed reduction are switched constantly, the car jolts is avoided, and the stability, reliability and comfort of car following are all improved.
Description
Technical field
The present invention relates to automobile active safety technical field, particularly a kind of intelligence is with vehicle control and method.
Background technology
Urban traffic blocking brings inconvenience to people's trip, brings serious social concern and environmental problem simultaneously.Such as, during traffic congestion, chaufeur needs the state constantly switching automobile between motoring condition and dead ship condition, causes chaufeur spirit high concentration, driving fatigue, improves the possibility that traffic accident occurs.
Intelligence can make chaufeur free from the driving fatigue during traffic congestion with the application of driving skills art, the generation avoided traffic accident.Current intelligence mainly adopts the device such as camera, radar to detect the actual spacing of vehicle and front vehicles with driving skills art, and adopt the mode of supervised learning, adjust the moving velocity of vehicle according to the difference of the following distance of actual spacing and setting, the deviation between actual spacing and the following distance of setting is minimized.
Realizing in process of the present invention, contriver finds that prior art at least exists following problem:
When actual spacing is greater than the following distance of setting, if front vehicles is due to emergency situations (as turning) just in time Reduced Speed Now, then accelerate the moving velocity of vehicle according to the difference of the following distance of actual spacing and setting, actual spacing can be caused to be less than again the following distance of setting.Now slow down according to the difference of the following distance of actual spacing and setting the moving velocity of vehicle again, if front vehicles (as completing turning) is given it the gun again, then not only can cause actual spacing but also be greater than the following distance of setting.So repeatedly, vehicle constantly accelerates and slows down, and causes vehicle to jolt, poor with the stability of car, reliability and traveling comfort.
Summary of the invention
Following the problem that the stability of car, reliability and traveling comfort are poor in order to solve prior art, embodiments providing a kind of intelligence with vehicle control and method.Described technical scheme is as follows:
On the one hand, embodiments provide a kind of intelligence with vehicle control, described intelligence comprises with vehicle control:
Detection module, for detecting the actual spacing of vehicle and front vehicles in real time, and produce the enhancing signal corresponding with described actual spacing according to described actual spacing, described enhancing signal for represent described actual spacing and setting dream car distance between deviation amplitude;
Strengthening study module, for adopting the mode strengthening study, according to described actual spacing and described enhancing signal, determining the adjusting range of the moving velocity of described vehicle;
With car module, for according to described adjusting range, adjust the moving velocity of described vehicle, to change described actual spacing.
In a kind of possible implementation of the present invention, described enhancing study module, comprising:
Action neural network, for according to described actual spacing, produces the adjusting range of the moving velocity of described vehicle;
Evaluate neural network, for according to described actual spacing, described enhancing signal and described adjusting range, produce cost function, described cost function is the approximate representation of described enhancing signal; According to described cost function, regulate the neural network weight of described evaluation neural network, to minimize the error of described cost function and described enhancing signal;
Described action neural network also for, according to the described cost function that the described evaluation neural network after neural network weight regulates produces, regulate the neural network weight of described action neural network, obtain optimum adjusting range, to minimize the error of described cost function and expectation value, described expectation value is the described cost function that described actual spacing produces when reaching described dream car distance.
Alternatively, described action neural network and described evaluation neural network all adopt Nonlinear Multi perceptron.
Alternatively, described action neural network is used for,
The neural network weight of Gradient Descent rule to described action neural network is adopted to carry out the adjustment of set point number;
Described evaluating network is used for,
The neural network weight of Gradient Descent rule to described evaluation neural network is adopted to carry out the adjustment of set point number.
In the another kind of possible implementation of the present invention, described adjusting range symbolization function.
On the other hand, embodiments provide a kind of intelligence with car control method, described intelligence comprises with car control method:
The actual spacing of real-time detection vehicle and front vehicles, and produce the enhancing signal corresponding with described actual spacing according to described actual spacing, described enhancing signal for represent described actual spacing and setting dream car distance between deviation amplitude;
Adopt the mode strengthening study, according to described actual spacing and described enhancing signal, determine the adjusting range of the moving velocity of described vehicle;
According to described adjusting range, adjust the moving velocity of described vehicle, to change described actual spacing.
In a kind of possible implementation of the present invention, described according to described actual spacing and described enhancing signal, determine the adjusting range of the moving velocity of described vehicle, comprising:
Adopt action neural network according to described actual spacing, produce the adjusting range of the moving velocity of described vehicle;
Adopt and evaluate neural network according to described actual spacing, described enhancing signal and described adjusting range, produce cost function, described cost function is the approximate representation of described enhancing signal;
According to described cost function, regulate the neural network weight of described evaluation neural network, to minimize the error of described cost function and described enhancing signal;
According to the described cost function that the described evaluation neural network after neural network weight regulates produces, regulate the neural network weight of described action neural network, obtain optimum adjusting range, to minimize the error of described cost function and expectation value, described expectation value is the described cost function that described actual spacing produces when reaching described dream car distance.
Alternatively, described action neural network and described evaluation neural network all adopt Nonlinear Multi perceptron.
Alternatively, described according to described cost function, regulate the neural network weight of described evaluation neural network, comprising:
The neural network weight of Gradient Descent rule to described action neural network is adopted to carry out the adjustment of set point number;
The described described cost function produced according to the described evaluation neural network after neural network weight adjustment, regulates the neural network weight of described action neural network, comprising:
The neural network weight of Gradient Descent rule to described evaluation neural network is adopted to carry out the adjustment of set point number.
In the another kind of possible implementation of the present invention, described adjusting range symbolization function.
The beneficial effect that the technical scheme that the embodiment of the present invention provides is brought is:
By adopting the mode strengthening study, according to actual spacing and enhancing signal, determine the adjusting range of the moving velocity of vehicle, strengthen signal for the dream car that represents actual spacing and setting apart between deviation amplitude, can in the adjustment process of moving velocity, adopt self adaptation dynamic programming method, constantly by strengthening the deviation amplitude of signal reflection, how autonomous learning determines suitable adjusting range according to actual spacing, effectively to adjust the moving velocity of vehicle, the deviation of the dream car distance of actual spacing and setting is minimized, the adjusting range of moving velocity only directly can not be determined according to the deviation amplitude between actual spacing and the dream car distance of setting, therefore there is not the situation constantly causing vehicle to jolt accelerating to switch between deceleration, with the stability of car, reliability and traveling comfort are all improved.
Accompanying drawing explanation
In order to be illustrated more clearly in the technical scheme in the embodiment of the present invention, below the accompanying drawing used required in describing embodiment is briefly described, apparently, accompanying drawing in the following describes is only some embodiments of the present invention, for those of ordinary skill in the art, under the prerequisite not paying creative work, other accompanying drawing can also be obtained according to these accompanying drawings.
Fig. 1 is the structural representation of a kind of intelligence of providing of the embodiment of the present invention one with vehicle control;
Fig. 2 is the structural representation of a kind of intelligence of providing of the embodiment of the present invention two with vehicle control;
Fig. 3 is the structural representation of the neural network that the embodiment of the present invention two provides;
Fig. 4 is the function curve diagram of the action neural network that the embodiment of the present invention two provides;
Fig. 5 is the diagram of circuit of a kind of intelligence of providing of the embodiment of the present invention three with car control method;
Fig. 6 is the diagram of circuit of a kind of intelligence of providing of the embodiment of the present invention four with car control method.
Detailed description of the invention
For making the object, technical solutions and advantages of the present invention clearly, below in conjunction with accompanying drawing, embodiment of the present invention is described further in detail.
Embodiment one
Embodiments provide a kind of intelligence with vehicle control, be specially adapted to the situation of low speed with car, see Fig. 1, this intelligence comprises with vehicle control:
Detection module 101, for detecting the actual spacing of vehicle and front vehicles in real time, and produces the enhancing signal corresponding with actual spacing according to actual spacing, strengthen signal for represent actual spacing and setting dream car distance between deviation amplitude;
Strengthening study module 102, for adopting the mode strengthening study, according to actual spacing and enhancing signal, determining the adjusting range of the moving velocity of vehicle;
With car module 103, for according to adjusting range, adjust the moving velocity of vehicle, to change actual spacing.
Easily know, strengthen and learn to be the Autonomous Agent (agent) by an energy perception environment, autonomous learning selects the optimum action that can reach its target.The process of concrete autonomous learning is that agency makes action in its environment, environment can feed back, agency's feedback (successfully award, unsuccessfully punish) environmentally, action is familiar with and learns, thus in follow-up action, pay the utmost attention to correct behavior and avoid the behavior of mistake to occur, unceasing study like this, finally can determine optimum action.In conjunction with the present embodiment, detection module 101 testing environment information (i.e. the actual spacing of vehicle and front vehicles), strengthen the first environmentally information of study module 102 and determine arbitrarily an action policy (i.e. the adjusting range of the moving velocity of vehicle), carry out action (namely adjusting the moving velocity of vehicle) with car module 103 according to the action policy determined.Then feed back (namely strengthen signal) again by testing environment information situation of change environmentally for detection module 101, strengthen study module 102 according to enhancing signal update action policy, carry out action with car module 103 according to the action policy after renewal.So constantly adjustment, until determine optimum action policy (namely actual spacing remains the dream car distance of setting).
Further, after carrying out action with car module 103, the actual spacing of vehicle and front vehicles can change, and the deviation amplitude between the dream car distance of actual spacing and setting may reduce, and also may increase.If deviation amplitude reduces, the punishment given by strengthening signal reduces thereupon, and strengthening study module 102 study is correct to action before, can preferentially adopt this action to carry out following car later; If deviation amplitude increases, the punishment given by strengthening signal increases thereupon, strengthens study module 102 and learns to action to be before wrong, can avoid adopting this action to carry out with car later.Particularly, when actual spacing is greater than the following distance of setting, first time is when causing actual spacing to be less than the following distance of setting due to front vehicles Reduced Speed Now, strengthen study module 102 to be punished accordingly by strengthening signal, just can avoid time secondary occurring identical situation as far as possible, decrease jolting of vehicle, improve with the stability of car, reliability and traveling comfort.
The embodiment of the present invention strengthens the mode of study by adopting, according to actual spacing and enhancing signal, determine the adjusting range of the moving velocity of vehicle, strengthen signal for the dream car that represents actual spacing and setting apart between deviation amplitude, can in the adjustment process of moving velocity, adopt self adaptation dynamic programming method, constantly by strengthening the deviation amplitude of signal reflection, how autonomous learning determines suitable adjusting range according to actual spacing, effectively to adjust the moving velocity of vehicle, the deviation of the dream car distance of actual spacing and setting is minimized, the adjusting range of moving velocity only directly can not be determined according to the deviation amplitude between actual spacing and the dream car distance of setting, therefore there is not the situation constantly causing vehicle to jolt accelerating to switch between deceleration, with the stability of car, reliability and traveling comfort are all improved.
Embodiment two
Embodiments provide a kind of intelligence with vehicle control, the present embodiment is the concrete discussion that the intelligence provided embodiment one follows vehicle control, and see Fig. 2, this intelligence comprises with vehicle control:
Detection module 201, for detecting the actual spacing of vehicle and front vehicles in real time, and produce the enhancing signal corresponding with actual spacing according to actual spacing, strengthen signal for representing that the dream car of actual spacing and setting is apart from the deviation amplitude between (as 15m);
Strengthening study module 202, for adopting the mode strengthening study, according to actual spacing and enhancing signal, determining the adjusting range of the moving velocity of vehicle;
With car module 203, for according to adjusting range, adjust the moving velocity of vehicle, to change actual spacing.
In the present embodiment, strengthen signal to set according to actual conditions.Deviation amplitude between the dream car distance of general actual spacing and setting is less, strengthens signal larger.Such as, when deviation amplitude is 0, strengthening signal is 0; When having collision danger or lose front vehicles, strengthening signal is-1.
In actual applications, detection module 201 can comprise:
Spacing acquiring unit 201a, for detecting the actual spacing of vehicle and front vehicles in real time;
Signal generation unit 201b, for the functional relation according to setting, by the spacing deviation between actual spacing and dream car distance, produces and strengthens signal.
Particularly, the functional relation of setting can be linear function, and the spacing deviation between actual spacing and dream car distance is larger, strengthens signal less.
In actual applications, acquiring unit 201a can be the sensor such as camera, radar (as millimeter wave radar, laser radar), and signal generation unit 201b can be micro controller system.
In a kind of implementation of the present embodiment, this enhancing study module 202 can comprise:
Action neural network 202a, for according to actual spacing, produces the adjusting range of the moving velocity of vehicle;
Evaluate neural network 202b, for according to actual spacing, enhancing signal and adjusting range, produce cost function, cost function is the approximate representation strengthening signal; According to cost function, regulate the neural network weight evaluating neural network 202b, to minimize cost function and the error strengthening signal;
Action neural network 202a also for, according to the cost function that the evaluation neural network 202b after neural network weight regulates produces, regulate the neural network weight of action neural network 202a, obtain optimum adjusting range, to minimize the error of cost function and expectation value, expectation value is the cost function that actual spacing produces when reaching dream car distance.
It should be noted that, the neural network weight of action neural network 202a, the neural network weight initial value evaluating neural network 202b can random arrangement, the process strengthening study module 202 study is mainly configuring initial value to the continuous adjustment of initial value, to the last obtain optimum weights, this time error reaches minimum.
Alternatively, action neural network 202a may be used for,
Gradient Descent rule is adopted the neural network weight of action neural network 202a to be carried out to the adjustment of set point number.
Alternatively, evaluate neural network 202b and may be used for,
Gradient Descent rule is adopted to carry out the adjustment of set point number to the neural network weight evaluating neural network 202b.
Understandably, adopt Gradient Descent rule to regulate, orderly adjustment can be realized on the one hand, improve the accuracy regulated, on the other hand can Step wise approximation, improve the efficiency regulated.
In specific implementation, first regulate the neural network weight evaluating neural network 202b according to set point number, to reduce cost function and the error strengthening signal; Then keep the neural network weight of the evaluation neural network 202b after upgrading, regulating the neural network weight of action neural network 202a according to set point number, with by changing adjusting range, minimizing the error of cost function and expectation value.And start to regulate the neural network weight evaluating neural network 202b ... backcycling like this, the neural network weight of to the last action neural network 202a, the neural network weight of evaluation neural network 202b all tend towards stability, the weights now obtained are optimum, and cost function reaches minimum, after this regulate according to these weights, just can reach the ideal distance of setting very soon.
Understandably, first regulate the neural network weight evaluating neural network 202b, cost function can be made more to approach enhancing signal, improve the accuracy evaluated; Regulate the neural network weight of action neural network 202a again, i.e. adjustable adjusting range (input value of the output valve of action neural network 202a and the neural network weight of evaluation neural network 202b), and then Readjusting cost function, reduce the deviation of itself and expectation value.
Particularly, evaluate neural network 202b according to following formula (1) calculation error, to have regulated when error is 0:
e
c(t)＝α*J(t)-[J(t-1)-r(t)]；(1)
Wherein, e
ct () represents cost function and the error strengthening signal, α is commutation factor, and 0 < α < 1, J (t) represents cost function, and r (t) represents enhancing signal, and t represents the moment.
Action neural network 202a according to following formula (2) calculation error, can regulate when error is 0:
e
a(t)＝J(t)-U(t)；(2)
Wherein, e
at () represents the error of cost function and expectation value, J (t) represents cost function, and U (t) represents expectation value, and t represents the moment.
In actual applications, by following formula (3), e (t) can be converted to E (t) to regulate:
E(t)＝(1/2)*[e(t)]
2；(3)
Wherein, e (t) is e
c(t) or e
a(t).
It should be noted that, adopting Gradient Descent rule to carry out adjustment is ask local derviation to weights, and e (t) is converted to E (t), can so that calculate.
Particularly, evaluate neural network 202b to regulate according to following formula (4)-(6):
w
c(t+1)＝w
c(t)+Δw
c(t)；(4)
Wherein, w
crepresent the input value evaluating neural network 202b, △ w
crepresent the intermediate result evaluating neural network 202b, l
crepresent the learning rate evaluating neural network 202b, E
crepresent the error evaluating neural network 202b, J represents cost function, and t represents the moment.
Action neural network 202a can regulate according to following formula (7)-(9):
w
a(t+1)＝w
a(t)+Δw
a(t)；(7)
Wherein, w
arepresent the input value of action neural network 202a, △ w
arepresent the intermediate result of action neural network 202a, l
arepresent the learning rate of action neural network 202a, E
arepresent the error of action neural network 202a, J represents cost function, and t represents the moment.
Preferably, action neural network 202a and evaluation neural network 202b can all adopt Nonlinear Multi perceptron.Multilayer perceptron comprises one or more layers hidden layer, and each hidden layer comprises some nodes, and each node of every layer exists mapping relations with each node of adjacent layer respectively.For one deck hidden layer shown in Fig. 3, x1-xn is n input value, and y1-ym is m node of hidden layer, and z is output valve, and xi is mapped to node yj with weight w ij, and the result that all input values calculate by node yj is mapped in output valve with weights vj.Wherein, n >=1 and n is integer, m >=1 and m is integer, 1≤i≤n and i is integer, 1≤j≤m and j is integer.In conjunction with action neural network 202a, input value is actual spacing, and output valve is adjusting range; Combining assessment neural network 202b, input value is actual spacing, strengthens signal, adjusting range, and output valve is cost function.
Easily know, the level of neural network is more, relation is more complicated, applicability and the accuracy of process event are higher, action neural network 202a and evaluation neural network 202b all adopts Nonlinear Multi perceptron, can improve the ability of action neural network 202a and evaluation neural network 202b autonomous learning, what autonomous learning was determined follows car strategy closer to optimum, with car better effects if.
In actual applications, detection module 201 can also comprise:
Normalization method unit 201c, for by actual spacing and strengthen signal normalization.
Alternatively, adjusting range can symbolization sgn function.
Such as, the physical relationship of adjusting range and actual spacing can as shown in Figure 4, and as can be seen from Figure 4, the span of adjusting range is [-1,1].Wherein, adjusting range is greater than 0 expression and needs open out, and 1 represents that throttle is added to maximum, and 0 represents without the need to open out or touches on the brake, and adjusting range is less than 0 expression needs and touches on the brake, and-1 represents that brake is stepped on maximum.
That is, the span of the input value of action neural network 202a and evaluation neural network 202b is [-1,1].In specific implementation, can first each input value be normalized, and then input action neural network 202a or evaluation neural network 202b.
Further, action neural network 202a and evaluation neural network 202b can adopt sigmoid function.
Understandably, all numerical value is normalized, and is more convenient for calculating, and is also convenient to make system be applicable to all vehicles.
In actual applications, the corresponding relation can trampling amplitude and operating range according to accelerator and brake specifically regulates, and can realize to make the present invention on various vehicle.
The embodiment of the present invention strengthens the mode of study by adopting, according to actual spacing and enhancing signal, determine the adjusting range of the moving velocity of vehicle, strengthen signal for the dream car that represents actual spacing and setting apart between deviation amplitude, can in the adjustment process of moving velocity, adopt self adaptation dynamic programming method, constantly by strengthening the deviation amplitude of signal reflection, how autonomous learning determines suitable adjusting range according to actual spacing, effectively to adjust the moving velocity of vehicle, the deviation of the dream car distance of actual spacing and setting is minimized, the adjusting range of moving velocity only directly can not be determined according to the deviation amplitude between actual spacing and the dream car distance of setting, therefore there is not the situation constantly causing vehicle to jolt accelerating to switch between deceleration, with the stability of car, reliability and traveling comfort are all improved.
Embodiment three
See Fig. 5, embodiments provide a kind of intelligence with car control method, be applicable to intelligence that embodiment one or embodiment two provide with vehicle control, the method comprises:
Step 301: the actual spacing detecting vehicle and front vehicles in real time, and produce the enhancing signal corresponding with actual spacing according to actual spacing.
In the present embodiment, strengthen signal for the dream car that represents actual spacing and setting apart between deviation amplitude.
Step 302: adopt the mode strengthening study, according to actual spacing and enhancing signal, determines the adjusting range of the moving velocity of vehicle.
Step 303: according to adjusting range, the moving velocity of adjustment vehicle, to change actual spacing.
The embodiment of the present invention strengthens the mode of study by adopting, according to actual spacing and enhancing signal, determine the adjusting range of the moving velocity of vehicle, strengthen signal for the dream car that represents actual spacing and setting apart between deviation amplitude, can in the adjustment process of moving velocity, adopt self adaptation dynamic programming method, constantly by strengthening the deviation amplitude of signal reflection, how autonomous learning determines suitable adjusting range according to actual spacing, effectively to adjust the moving velocity of vehicle, the deviation of the dream car distance of actual spacing and setting is minimized, the adjusting range of moving velocity only directly can not be determined according to the deviation amplitude between actual spacing and the dream car distance of setting, therefore there is not the situation constantly causing vehicle to jolt accelerating to switch between deceleration, with the stability of car, reliability and traveling comfort are all improved.
Embodiment four
See Fig. 6, embodiments provide a kind of intelligence with car control method, be applicable to intelligence that embodiment one or embodiment two provide with vehicle control, the present embodiment is the concrete discussion with car control method of the intelligence that provides embodiment three, and the method comprises:
Step 401: the actual spacing detecting vehicle and front vehicles in real time, and produce the enhancing signal corresponding with actual spacing according to actual spacing.
In the present embodiment, strengthen signal for the dream car that represents actual spacing and setting apart between deviation amplitude.
Step 402: adopt the mode strengthening study, according to actual spacing and enhancing signal, determines the adjusting range of the moving velocity of vehicle.
In a kind of implementation of embodiment, this step 402 can comprise:
Step 402a: adopt action neural network according to actual spacing, produce the adjusting range of the moving velocity of vehicle.
Step 402b: adopt and evaluate neural network according to actual spacing, enhancing signal and adjusting range, produce cost function.
In the present embodiment, cost function is the approximate representation strengthening signal.
Step 402c: according to cost function, regulates the neural network weight evaluating neural network, to minimize cost function and the error strengthening signal.
Alternatively, this step 402c can comprise:
Gradient Descent rule is adopted the neural network weight of action neural network to be carried out to the adjustment of set point number.
Step 402d: the cost function produced according to the evaluation neural network after neural network weight regulates, regulates the neural network weight of action neural network, obtain optimum adjusting range, to minimize the error of cost function and expectation value.
In the present embodiment, expectation value is the cost function that actual spacing produces when reaching dream car distance.
Alternatively, this step 402d can comprise:
Gradient Descent rule is adopted to carry out the adjustment of set point number to the neural network weight evaluating neural network.
Preferably, action neural network and evaluation neural network can all adopt Nonlinear Multi perceptron.
Step 403: according to adjusting range, the moving velocity of adjustment vehicle, to change actual spacing.
Alternatively, adjusting range can symbolization function.
The embodiment of the present invention strengthens the mode of study by adopting, according to actual spacing and enhancing signal, determine the adjusting range of the moving velocity of vehicle, strengthen signal for the dream car that represents actual spacing and setting apart between deviation amplitude, can in the adjustment process of moving velocity, adopt self adaptation dynamic programming method, constantly by strengthening the deviation amplitude of signal reflection, how autonomous learning determines suitable adjusting range according to actual spacing, effectively to adjust the moving velocity of vehicle, the deviation of the dream car distance of actual spacing and setting is minimized, the adjusting range of moving velocity only directly can not be determined according to the deviation amplitude between actual spacing and the dream car distance of setting, therefore there is not the situation constantly causing vehicle to jolt accelerating to switch between deceleration, with the stability of car, reliability and traveling comfort are all improved.
It should be noted that: the intelligence that above-described embodiment provides follows vehicle control when controlling intelligence with car, only be illustrated with the division of above-mentioned each functional module, in practical application, can distribute as required and by above-mentioned functions and be completed by different functional modules, inner structure by system is divided into different functional modules, to complete all or part of function described above.In addition, the intelligence that above-described embodiment provides belongs to same design with vehicle control and intelligence with car control method embodiment, and its specific implementation process refers to embodiment of the method, repeats no more here.
The invention described above embodiment sequence number, just to describing, does not represent the quality of embodiment.
One of ordinary skill in the art will appreciate that all or part of step realizing above-described embodiment can have been come by hardware, the hardware that also can carry out instruction relevant by program completes, described program can be stored in a kind of computer-readable recording medium, the above-mentioned storage medium mentioned can be read-only memory (ROM), disk or CD etc.
The foregoing is only preferred embodiment of the present invention, not in order to limit the present invention, within the spirit and principles in the present invention all, any amendment done, equivalent replacement, improvement etc., all should be included within protection scope of the present invention.
Claims (10)
1. intelligence is with a vehicle control, it is characterized in that, described intelligence comprises with vehicle control:
Detection module, for detecting the actual spacing of vehicle and front vehicles in real time, and produce the enhancing signal corresponding with described actual spacing according to described actual spacing, described enhancing signal for represent described actual spacing and setting dream car distance between deviation amplitude;
Strengthening study module, for adopting the mode strengthening study, according to described actual spacing and described enhancing signal, determining the adjusting range of the moving velocity of described vehicle;
With car module, for according to described adjusting range, adjust the moving velocity of described vehicle, to change described actual spacing.
2. intelligence according to claim 1 is with vehicle control, and it is characterized in that, described enhancing study module, comprising:
Action neural network, for according to described actual spacing, produces the adjusting range of the moving velocity of described vehicle;
Evaluate neural network, for according to described actual spacing, described enhancing signal and described adjusting range, produce cost function, described cost function is the approximate representation of described enhancing signal; According to described cost function, regulate the neural network weight of described evaluation neural network, to minimize the error of described cost function and described enhancing signal;
Described action neural network also for, according to the described cost function that the described evaluation neural network after neural network weight regulates produces, regulate the neural network weight of described action neural network, obtain optimum adjusting range, to minimize the error of described cost function and expectation value, described expectation value is the described cost function that described actual spacing produces when reaching described dream car distance.
3. intelligence according to claim 2 is with vehicle control, and it is characterized in that, described action neural network and described evaluation neural network all adopt Nonlinear Multi perceptron.
4. intelligence according to claim 2 is with vehicle control, and it is characterized in that, described action neural network is used for,
The neural network weight of Gradient Descent rule to described action neural network is adopted to carry out the adjustment of set point number;
Described evaluating network is used for,
The neural network weight of Gradient Descent rule to described evaluation neural network is adopted to carry out the adjustment of set point number.
5. the intelligence according to any one of claim 1-4, with vehicle control, is characterized in that, described adjusting range symbolization function.
6. intelligence is with a car control method, it is characterized in that, described intelligence comprises with car control method:
The actual spacing of real-time detection vehicle and front vehicles, and produce the enhancing signal corresponding with described actual spacing according to described actual spacing, described enhancing signal for represent described actual spacing and setting dream car distance between deviation amplitude;
Adopt the mode strengthening study, according to described actual spacing and described enhancing signal, determine the adjusting range of the moving velocity of described vehicle;
According to described adjusting range, adjust the moving velocity of described vehicle, to change described actual spacing.
7. intelligence according to claim 6 is with car control method, it is characterized in that, described according to described actual spacing and described enhancing signal, determines the adjusting range of the moving velocity of described vehicle, comprising:
Adopt action neural network according to described actual spacing, produce the adjusting range of the moving velocity of described vehicle;
Adopt and evaluate neural network according to described actual spacing, described enhancing signal and described adjusting range, produce cost function, described cost function is the approximate representation of described enhancing signal;
According to described cost function, regulate the neural network weight of described evaluation neural network, to minimize the error of described cost function and described enhancing signal;
According to the described cost function that the described evaluation neural network after neural network weight regulates produces, regulate the neural network weight of described action neural network, obtain optimum adjusting range, to minimize the error of described cost function and expectation value, described expectation value is the described cost function that described actual spacing produces when reaching described dream car distance.
8. intelligence according to claim 7 is with car control method, and it is characterized in that, described action neural network and described evaluation neural network all adopt Nonlinear Multi perceptron.
9. intelligence according to claim 7 is with car control method, it is characterized in that, described according to described cost function, regulates the neural network weight of described evaluation neural network, comprising:
The neural network weight of Gradient Descent rule to described action neural network is adopted to carry out the adjustment of set point number;
The described described cost function produced according to the described evaluation neural network after neural network weight adjustment, regulates the neural network weight of described action neural network, comprising:
The neural network weight of Gradient Descent rule to described evaluation neural network is adopted to carry out the adjustment of set point number.
10. the intelligence according to any one of claim 6-9, with car control method, is characterized in that, described adjusting range symbolization function.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510489170.5A CN105059213B (en) | 2015-08-11 | 2015-08-11 | A kind of intelligence is with vehicle control and method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510489170.5A CN105059213B (en) | 2015-08-11 | 2015-08-11 | A kind of intelligence is with vehicle control and method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105059213A true CN105059213A (en) | 2015-11-18 |
| CN105059213B CN105059213B (en) | 2017-08-04 |
Family
ID=54488819
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510489170.5A Active CN105059213B (en) | 2015-08-11 | 2015-08-11 | A kind of intelligence is with vehicle control and method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105059213B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105835854A (en) * | 2016-03-17 | 2016-08-10 | 奇瑞汽车股份有限公司 | Emergency braking control system and control method thereof |
| CN107065561A (en) * | 2017-05-16 | 2017-08-18 | 清华大学 | The machine learning control method of double-wheel self-balancing car |
| CN109686086A (en) * | 2018-12-24 | 2019-04-26 | 东软集团（北京）有限公司 | The training of fuzzy control network generates the method and device that speed is suggested at crossing |
| CN109849911A (en) * | 2019-02-19 | 2019-06-07 | 百度在线网络技术（北京）有限公司 | Follow the bus method, apparatus and computer readable storage medium |
| CN110194156A (en) * | 2019-06-21 | 2019-09-03 | 厦门大学 | Intelligent network joins hybrid vehicle active collision avoidance enhancing learning control system and method |
| CN112100855A (en) * | 2020-09-17 | 2020-12-18 | 北京智能车联产业创新中心有限公司 | Vehicle following capability evaluation method and device, electronic device and storage medium |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH08127268A (en) * | 1994-10-28 | 1996-05-21 | Isuzu Motors Ltd | Vehicle running controller |
| CN101417655A (en) * | 2008-10-14 | 2009-04-29 | 清华大学 | Vehicle multi-objective coordinated self-adapting cruise control method |
| JP2009149173A (en) * | 2007-12-19 | 2009-07-09 | Mitsubishi Fuso Truck & Bus Corp | Auto-cruise device |
| JP2011095834A (en) * | 2009-10-27 | 2011-05-12 | Denso Corp | Device and program for controlling vehicle |
| CN103129556A (en) * | 2011-11-22 | 2013-06-05 | 罗伯特·博世有限公司 | Driving assistance system |
| KR20130059702A (en) * | 2011-11-29 | 2013-06-07 | 현대자동차주식회사 | Apparatus and method for controlling obstacle sensing zone |
| CN104024021A (en) * | 2012-01-02 | 2014-09-03 | 沃尔沃拉斯特瓦格纳公司 | Method and system for controlling a driving distance |
| CN104554264A (en) * | 2014-11-28 | 2015-04-29 | 温州大学 | Method and system for self-adaptively on-line adjusting vehicle speed |
-
2015
- 2015-08-11 CN CN201510489170.5A patent/CN105059213B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH08127268A (en) * | 1994-10-28 | 1996-05-21 | Isuzu Motors Ltd | Vehicle running controller |
| JP2009149173A (en) * | 2007-12-19 | 2009-07-09 | Mitsubishi Fuso Truck & Bus Corp | Auto-cruise device |
| CN101417655A (en) * | 2008-10-14 | 2009-04-29 | 清华大学 | Vehicle multi-objective coordinated self-adapting cruise control method |
| JP2011095834A (en) * | 2009-10-27 | 2011-05-12 | Denso Corp | Device and program for controlling vehicle |
| CN103129556A (en) * | 2011-11-22 | 2013-06-05 | 罗伯特·博世有限公司 | Driving assistance system |
| KR20130059702A (en) * | 2011-11-29 | 2013-06-07 | 현대자동차주식회사 | Apparatus and method for controlling obstacle sensing zone |
| CN104024021A (en) * | 2012-01-02 | 2014-09-03 | 沃尔沃拉斯特瓦格纳公司 | Method and system for controlling a driving distance |
| CN104554264A (en) * | 2014-11-28 | 2015-04-29 | 温州大学 | Method and system for self-adaptively on-line adjusting vehicle speed |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105835854A (en) * | 2016-03-17 | 2016-08-10 | 奇瑞汽车股份有限公司 | Emergency braking control system and control method thereof |
| CN105835854B (en) * | 2016-03-17 | 2018-11-16 | 奇瑞汽车股份有限公司 | A kind of emergency braking control system and its control method |
| CN107065561A (en) * | 2017-05-16 | 2017-08-18 | 清华大学 | The machine learning control method of double-wheel self-balancing car |
| CN107065561B (en) * | 2017-05-16 | 2019-11-22 | 清华大学 | The machine learning control method of double-wheel self-balancing vehicle |
| CN109686086A (en) * | 2018-12-24 | 2019-04-26 | 东软集团（北京）有限公司 | The training of fuzzy control network generates the method and device that speed is suggested at crossing |
| CN109849911A (en) * | 2019-02-19 | 2019-06-07 | 百度在线网络技术（北京）有限公司 | Follow the bus method, apparatus and computer readable storage medium |
| CN110194156A (en) * | 2019-06-21 | 2019-09-03 | 厦门大学 | Intelligent network joins hybrid vehicle active collision avoidance enhancing learning control system and method |
| CN110194156B (en) * | 2019-06-21 | 2020-11-10 | 厦门大学 | Intelligent network-connected hybrid electric vehicle active collision avoidance reinforcement learning control system and method |
| CN112100855A (en) * | 2020-09-17 | 2020-12-18 | 北京智能车联产业创新中心有限公司 | Vehicle following capability evaluation method and device, electronic device and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105059213B (en) | 2017-08-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105059213A (en) | Intelligent car following control system and method | |
| CN111898211B (en) | Intelligent vehicle speed decision method based on deep reinforcement learning and simulation method thereof | |
| Ntousakis et al. | On microscopic modelling of adaptive cruise control systems | |
| CN105059288A (en) | Lane keeping control system and method | |
| CN107168303A (en) | A kind of automatic Pilot method and device of automobile | |
| Cai et al. | An intelligent longitudinal controller for application in semiautonomous vehicles | |
| CN105035085A (en) | Method and device for automatic car following | |
| CN105109482A (en) | Parking garage entry method and device | |
| US20200310420A1 (en) | System and method to train and select a best solution in a dynamical system | |
| CN105109488B (en) | A kind of intelligence is with car system and method | |
| Zhang et al. | A cruise control method for connected vehicle systems considering side vehicles merging behavior | |
| EP3594770A1 (en) | Deep reinforcement learning for a general framework for model-based longitudinal control | |
| Tajdari et al. | Fuzzy control of anticipation and evaluation behaviour in real traffic flow | |
| Chin et al. | Model predictive cooperative cruise control in mixed traffic | |
| Zhao et al. | Coordinated throttle and brake fuzzy controller design for vehicle following | |
| Cherian et al. | Neural network based ACC for optimized safety and comfort | |
| Abdullahi et al. | Adaptive cruise control: A model reference adaptive control approach | |
| Selvaraj et al. | An ML-aided reinforcement learning approach for challenging vehicle maneuvers | |
| Xu et al. | Hierarchical speed control for autonomous electric vehicle through deep reinforcement learning and robust control | |
| Zhao et al. | Supervised adaptive dynamic programming based adaptive cruise control | |
| Seo et al. | Robust mode predictive control for lane change of automated driving vehicles | |
| Li et al. | Traffic-aware ecological cruising control for connected electric vehicle | |
| Li et al. | Energy-efficient autonomous vehicle control using reinforcement learning and interactive traffic simulations | |
| Sulejmani et al. | Autonomous overtaking assistant for country road scenarios | |
| CN109144107B (en) | Automatic car following control method and device and car |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20220218 Address after: 241006 Anshan South Road, Wuhu Economic and Technological Development Zone, Anhui Province Patentee after: Wuhu Sambalion auto technology Co.,Ltd. Address before: 241006 Changchun Road, Wuhu economic and Technological Development Zone, Wuhu, Anhui, 8 Patentee before: CHERY AUTOMOBILE Co.,Ltd. |
|
| TR01 | Transfer of patent right |