CN104760592B - Self-propelled vehicle Decision Control method - Google Patents

Self-propelled vehicle Decision Control method Download PDF

Info

Publication number
CN104760592B
CN104760592B CN201510081698.9A CN201510081698A CN104760592B CN 104760592 B CN104760592 B CN 104760592B CN 201510081698 A CN201510081698 A CN 201510081698A CN 104760592 B CN104760592 B CN 104760592B
Authority
CN
China
Prior art keywords
maneuver
self
virtual world
propelled vehicle
information
Prior art date
Application number
CN201510081698.9A
Other languages
Chinese (zh)
Other versions
CN104760592A (en
Inventor
潘晨劲
赵江宜
Original Assignee
福州华鹰重工机械有限公司
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 福州华鹰重工机械有限公司 filed Critical 福州华鹰重工机械有限公司
Priority to CN201510081698.9A priority Critical patent/CN104760592B/en
Publication of CN104760592A publication Critical patent/CN104760592A/en
Application granted granted Critical
Publication of CN104760592B publication Critical patent/CN104760592B/en

Links

Classifications

    • BPERFORMING OPERATIONS; TRANSPORTING
    • B60VEHICLES IN GENERAL
    • B60WCONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
    • B60W50/00Details of control systems for road vehicle drive control not related to the control of a particular sub-unit, e.g. process diagnostic or vehicle driver interfaces
    • B60W50/0097Predicting future conditions
    • BPERFORMING OPERATIONS; TRANSPORTING
    • B60VEHICLES IN GENERAL
    • B60WCONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
    • B60W2555/00Input parameters relating to exterior conditions, not covered by groups B60W2552/00, B60W2554/00
    • B60W2555/60Traffic rules, e.g. speed limits or right of way
    • BPERFORMING OPERATIONS; TRANSPORTING
    • B60VEHICLES IN GENERAL
    • B60WCONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
    • B60W2556/00Input parameters relating to data
    • B60W2556/45External transmission of data to or from the vehicle
    • B60W2556/50External transmission of data to or from the vehicle for navigation systems
    • BPERFORMING OPERATIONS; TRANSPORTING
    • B60VEHICLES IN GENERAL
    • B60WCONJOINT CONTROL OF VEHICLE SUB-UNITS OF DIFFERENT TYPE OR DIFFERENT FUNCTION; CONTROL SYSTEMS SPECIALLY ADAPTED FOR HYBRID VEHICLES; ROAD VEHICLE DRIVE CONTROL SYSTEMS FOR PURPOSES NOT RELATED TO THE CONTROL OF A PARTICULAR SUB-UNIT
    • B60W2556/00Input parameters relating to data
    • B60W2556/45External transmission of data to or from the vehicle
    • B60W2556/65Data transmitted between vehicles

Abstract

The open self-propelled vehicle Decision Control method of the present invention, comprises the steps: the event information obtaining the input of virtual world module; Obtain the routing information of navigation module input; According to event information and routing information generation machine motor-car maneuver.The technical solution of the present invention integrated virtual world and navigation module, make the control of self-propelled vehicle meet actual demand for control, and can generate safe control policy.

Description

Self-propelled vehicle Decision Control method

Technical field

The present invention relates to self-propelled vehicle control technology field, particularly relate to self-propelled vehicle Decision Control method.

Background technology

Behaviour decision making module (namely judge the environment of self-propelled vehicle and carry out decision-making according to judged result to the behavior of self-propelled vehicle) is responsible for high-rise wagon control task, relative to the bottom wagon control task that action decision-making module is responsible for.The main task of high-rise wagon control is Real-time Decision and generates, monitors, controls various driving maneuver action (hereinafter referred to as maneuver).And the main task of bottom wagon control is responsible for various maneuver to be converted into the instruction of fax manipulation or brake, throttle, bearing circle, change speed gear box driving machine.

Maneuver has locality feature clearly, in certain hour and certain limit for reaching a motion of certain object.The object of maneuver is, while the various situations of encountering in reply is driven, close to driving destination.As the driving experience of the mankind, the driving procedure of (A, B can be any two points in urban road network) can be decomposed into the maneuver of some mutual linkings from point A to point B.Such as, the driving procedure from the garage of family to office building parking area can comprise these continuous print maneuvers:

1. roll garage away from

2. incorporate garage door a kind of horse raised north of the Grean Wall road

3. travel along road

4. overtake other vehicles

5. travel along road

6. cross crossing

7. follow front vehicles to travel

8. cross crossing

9. emergency braking

10. travel along road

11. enter parking area

12. stop into parking stall.

Owing to facing road conditions and the environmental aspect of constantly change in driving procedure, behaviour decision making module needs to grasp exactly most news, immediately makes a policy, and performs an optimum maneuver.

Further, behaviour decision making module stops according to new situation the maneuver that performing at any time possibly, and the maneuver that Tactic selection one is new again performs.Such as, in the process of following front vehicles traveling, front vehicles is slowed down suddenly, and this time will be changed into " emergency braking " by " following front vehicles ").Namely self-propelled vehicle is needed to realize the Decision Control to self-propelled vehicle according to different ambient conditions.

Summary of the invention

For this reason, need to provide self-propelled vehicle Decision Control scheme, solve the Decision Control problem of self-propelled vehicle in different scene.

For achieving the above object, inventor provide self-propelled vehicle Decision Control method, be applied to the behaviour decision making module of self-propelled vehicle, comprise the steps:

Obtain the event information of virtual world module input;

Obtain the routing information of navigation module input;

According to event information and routing information generation machine motor-car maneuver.

Further, also comprise the steps:

When the road conditions of the information on electronic chart and reality are inconsistent, export traffic information to navigation module.

Further, also comprise the steps:

The shape of road of reappraising and trend;

Export road information to guidance module.

Further, described event information comprises virtual world event, and described virtual world event comprises:

The virtual world event of obstacle class, the virtual world event of vehicle oneself state, the virtual world event of vehicle current driving road segment or the virtual world event of perception aspect.

Further, analyze the routing information of navigation module input, therefrom extract the route planning of motor-driven vehicle going;

Wherein, route planning comprise keep straight on forward, turn left forward, turn right forward or turn around travel.

Further, described " generation machine motor-car maneuver " comprises according to finite state automaton generation machine motor-car maneuver.

Further, described " generation machine motor-car maneuver " comprises the self-propelled vehicle maneuver that use Petri net generates safety.

Further, described " generation machine motor-car maneuver " comprises the steps:

According to event information and traffic regulation and drive the maneuver autopilot method that knowledge information generates safety;

Safe and feasible is generated and maneuver autopilot method reconcilable with route planning according to the maneuver autopilot method generated and route planning subnet information.

Further, also step is comprised:

When receiving error condition;

Judge whether the Fault recovery maneuver autopilot algorithm meeting this error condition, have, perform this Fault recovery maneuver autopilot algorithm;

Otherwise, start the performance period of a new round, according to new status information, select the maneuver autopilot algorithm that is new.

Be different from prior art, the technique scheme integrated virtual world and navigation module, make the control of self-propelled vehicle meet actual demand for control, and safe control policy can be generated.

Accompanying drawing explanation

Fig. 1 is the interface schematic diagram that behaviour decision making module exchanges with action control module;

Fig. 2 is the finite state automaton schematic diagram of self-propelled vehicle control algorithm implementation;

Fig. 3 is the decision process schematic diagram of a use Petri net Simulation of Complex;

Fig. 4 is a Petri net network diagram;

Fig. 5 is the schematic diagram of the state variation rule of Petri net;

Fig. 6 is the prototype structure schematic diagram of some Petri nets;

Fig. 7 is the schematic diagram of decision-making mode of multi input, multi output structure;

Fig. 8 is single input, list exports, the structural representation of the decision-making mode of multistage transition;

Fig. 9 is the structural representation of the self-propelled vehicle Decision Control method of a decision-making mode local.

Detailed description of the invention

By describe in detail technical scheme technology contents, structural attitude, realized object and effect, coordinate accompanying drawing to be explained in detail below in conjunction with specific embodiment.

Refer to shown in Fig. 1 to Fig. 9, the present embodiment provides self-propelled vehicle Decision Control method, is applied to the behaviour decision making module of self-propelled vehicle, comprises the steps: the event information obtaining the input of virtual world module; Obtain the routing information of navigation module input; According to event information and routing information generation machine motor-car maneuver.Wherein, the input Data Source of behaviour decision making module has: virtual world module in sensing module: virtual world with the form of virtual world event sets tuple for behaviour decision making module provides event information relevant with Driving Decision-making accurately, timely.Navigation module: export an optimal path, optimal path tells the Main way of behavior decision-making module current driving and the selection in track.Behaviour decision making module can also return navigation module by feedback information in some cases.

As shown in Figure 1, the interface that behaviour decision making module exchanges with action control module is performance-based objective instruction.Action target instruction target word in Fig. 1, each action target instruction target word:

● comprise an object point.

O object point usually within local scope, i.e. within effective sensing range of sensor.

O object point vectoring aircraft motor-car is close/arrive next way point.

● be divided into elementary instruction and extended instruction.

Performance-based objective instruction transfer to action control module process (can REFERENCE TO RELATED people application the patent No.: the Chinese patent application of 201510005390.6) after, generate path of motion by action control module, and this path of motion can be realized, the control signal that directly can be performed by self-propelled vehicle fax control system.

The optimal path information that navigation module generates is accepted when behaviour decision making module major part.In some cases, behaviour decision making module can output information to navigation module: as in actual driving procedure, find that the information on electronic chart is inconsistent with the road conditions of reality, such as: because traffic accident, roadupkeep etc. cause a transitable road originally to become impassabitity; Or there is error in electronic chart, inconsistent on the geometric configuration of real road and topological structure and electronic chart.When the road conditions of the information namely on electronic chart and reality are inconsistent, behaviour decision making module can export traffic information to navigation module.

This time, behaviour decision making module meeting: open road mapping algorithm, the shape of road of reappraising and trend.This method can ensure that self-propelled vehicle has the ability to explore unknown road, until come back on the road identical with electronic chart.While exploring road with road mapping algorithm, constantly new road information is fed back to navigation module.Navigation module can upgrade electronic chart according to this.The maximum duration explored under unknown road condition is determined by navigation module.If explore under unknown road condition and exceeded this time, then Behavior-Based control module can be abandoned continuing to explore, and present road is labeled as impassabitity, and requires that navigation module regenerates an optimal path.

In sensing module, include following information:

1. the information (information gathered in advance) of the priori in virtual world is stored in electronic chart.Which includes, the information that the lane on road network, road geometry, road is relevant with driving with Lu Jie, traffic lights three-dimensional position and classification etc.

2. simultaneously, the real time data that onboard sensor gathers, through various process, constantly can upgrade the information in the local category in virtual world.

3. that simultaneously obtain from in-vehicle communication system, also constantly can upgrade virtual world and sensing module from the information of other online vehicles or urban traffic network state.

The Data Update real-time each time described in above 2nd and the 3rd situation, all likely triggers the generation of virtual world event.

Be responsible for the realization of the software module producing virtual world event, can by arranging function hook (functionhook) in virtual world software module.

Function hook (hook) is arranged on certain function of code channel (codepath) of Data Update.

On the code channel of each Data Update, all will hook be set, to ensure that any Data Update that may produce virtual world event all can not be missed.

Function hook points to one section of program (calling virtual world event generator in the following text), and this section of program is responsible for judging whether to trigger some (one or more) virtual world event.

Virtual world event generator can access the particular content that current data upgrades, also can all information in accesses virtual world software module.

Each virtual world event be one predefined, reflect some situations relevant with Driving Decision-making in surrounding environment.

Virtual world event generator, all can according to current new real time data when being triggered each time, and the existing various information of virtual world, judges whether the condition meeting certain virtual world event is set up.The concrete foundation judged can be:

● does sensor obtain some specific information within the scope of certain reliability?

● do some specific conditions meet in virtual world?

● judge in real world, to there occurs some specific events from more new data?

Virtual world event can comprise according to classification:

● the virtual world event of obstacle class:

O vehicle front n rice has pedestrian.

O vehicle left side track is traveling lane in the same way.

Obstacle is not had in the positive left-hand lane of o vehicle.

Mobile (safety distance depends on current maneuver vehicle speed) is not had in o vehicle left side track, in the safety distance of left back.

O vehicle left side track is occupied by other mobiles.

● the virtual world event of vehicle oneself state:

O vehicle-mounted global location/inertial navigation is working properly.

The parameter of o pressure of tire sensor is normal.

The parameter of o fuel sensor is normal.

The parameter of o engine temperature sensor is normal.

The parameter of o braking mode sensor is normal.

● the virtual world event of vehicle current driving road segment:

O vehicle is just travelling on a highway.

O vehicle is just travelling on urban highway.

O vehicle front 100 meters of right turn lanes are first, second track (from the number of the right).

O vehicle front 30 meters has zebra crossing.

● the virtual world event of perception aspect

60 degree, o dead ahead is stopped by large obstacle in the visual field.

O junction ahead vehicle condition in the n rice of track of turning right is failed to understand (because the information that the visual field is stopped or sensor provides is not enough to do clear and definite judgement)

As can be seen here, each virtual world event will have certain atomicity (atomicity), conveniently the structure of decision-making mode in decision system.

The set (hereinafter referred to as virtual world event sets) of all virtual world events will contain human brain in the process of Driving Decision-making, the environmental element of required concern as far as possible.

Virtual world event sets can comprise abundant semanteme, thus fully states the various situations faced in driving procedure.

Virtual world event sets can be expressed with a tuple (tuple):

E=(e 1,e 2,…,e k)

Wherein each e l(l=1,2 ..., k) represent a specific virtual world event

Each specific virtual world event e lcan have two states, true (this virtual world event there occurs/condition meet) or false (this virtual world event do not occur/condition do not meet), with mathematical linguistics expression and e l∈ { true, false}.

Virtual world event generator upgrades virtual world event sets each time, and the virtual world event assignment that all current all conditions can be met is true, and condition unsatisfied virtual world event assignment is false.

Virtual world event sets also with a timestamp (timestamp), provides ageing information simultaneously.

Virtual world event sets tuple provides the structure that self-propelled vehicle Decision Control method flexibly, is very easily expanded.The classification of the virtual world event in set can include but not limited to above exemplified classification, can include but not limited to above exemplified concrete virtual world event in each classification.Different virtual events can be defined as required, in order to as far as possible intactly cover the varying environment that current application faces in practical application.

Such as, rational virtual world event sets can suitably be formulated according to the driving rule in country variant area and road conditions.

Again such as, and the feature in the region that mainly can travel according to automatic Pilot self-propelled vehicle (as southern hills highway, municipal highway, paddy field, south net highway, the highway of northeast possibility accumulated snow, municipal highway, township road etc.) formulate rational virtual world event sets.

The terseness at this interface of virtual world event sets tuple is conducive to the information exchange with other module/submodules.

Before personal vehicle system comes into operation, need user to provide and travel destination.Travel after destination obtaining, navigation module, with reference to electronic chart (can the application number of REFERENCE TO RELATED people application be: the Chinese patent application of 201510040520.X), generates the optimal route of an arrival destination.This route determines one group of way point that must arrive in order.

Under many situations, the selection of route has important impact for current Driving Decision-making, such as following situation:

To turn right according to the next crossing of route planning.The next crossing of current maneuver spacing not far (such as within 100 meters).Self-propelled vehicle is positioned at now right road and travels.In same track, self-propelled vehicle dead ahead, nearby (in 20 meters) have a slow-paced vehicle in traveling.In this case, because predict the route planning next will turned right, behaviour decision making module is reasonably selected should be and is continued to remain on current lane traveling, to facilitate ensuing right-hand rotation.If the information of no future route planning when doing Driving Decision-making, then behaviour decision making module selects lane change to surmount the vehicle of front obstruction possibly.But this selection probably causes self-propelled vehicle be in non-right turn lane when arriving crossing and need to perform relatively fierce action right-hand rotation.

Again such as, if current route plan is " turning around to travel ", then maneuver of now overtaking other vehicles is not just one and reasonably selects.

Therefore, in order to make the decision-making made forward-looking, the best route that behaviour decision making module Water demand navigation module generates, therefrom extracts the route planning next travelled.

Route planning is at any one time all the some elements in following set:

D={ keeps straight on forward, turns left forward, turns right forward, turns around to travel }

Wherein,

● keep straight on forward and represent that continuation is advanced a period of time along present road by vehicle.

● turn left forward, turning right forward represents that vehicle will be turned left or turn right at short notice.

● it is motor-driven that the traveling that turns around represents that vehicle needs to do U-shaped at once.

The unit be made up of the element in set D consists of route planning tuple.Similar with virtual world event sets tuple, each the element value in route planning tuple only has true/false two kinds of selections.Each moment can only have the value of an element to be true in addition.

The needs that route planning in D set can realize according to reality, do suitable adjustment.

Meanwhile, the extraction of route planning needs certain lead.Lead specifically determines (the rational scope of lead can be 50-150 rice) according to factors such as current driving environment, the speed of a motor vehicle.

Do decision-making each time, behaviour decision making module all needs a current route to be intended to be the input data (input) of decision-making.

The extraction of route planning can by one independently software submodules be responsible for, also can by a sub-module in charge in navigation module or behaviour decision making module.

Driving maneuver action control algorithm (hereinafter referred to as maneuver autopilot algorithm) is responsible for controlling machine motor-car and is made a series of maneuver within the specific limits with in certain hour.As mentioned above, the object of these maneuvers is: while the various situations of encountering in reply is driven, close to driving destination.

Behaviour decision making module has pre-established one group of maneuver autopilot algorithm, defines the specific purposes of each maneuver autopilot algorithm, strategy, step, maneuver.Because the implementation of all maneuver autopilot algorithms can be simulated (as described in Figure 2) with a finite state automaton (finitestateautomata), so behaviour decision making module also specify executing state number, State Transferring equation, the State Transferring event and state maintenance condition of the automat of each maneuver autopilot algorithm.

Each maneuver autopilot algorithm is a Closed control algorithm (closed-loopalgorithm), i.e. controlled object (self-propelled vehicle) output after accepting instruction (status information after actual execution) can feed back in control algorithm, and has influence on the generation of the next control command of control algorithm.

Below enumerate several modal maneuver autopilot algorithm.Certain maneuver autopilot algorithm is never only limitted to that these are several, but can formulate according to the actual needs.More maneuver autopilot algorithm can make the action of self-propelled vehicle more flexible, and the selection under same road conditions is more.

● arrive coordinate points (travelling along road)

● cross crossing

● overtake other vehicles

● follow front vehicles

● emergency braking

Fig. 2 is the finite state automaton schematic diagram of maneuver autopilot algorithm implementation.As seen from Figure 2, the implementation of maneuver autopilot algorithm forms (q by several different executing states 0, q 1..., q n).Conversion between executing state is triggered by executing state change event.Maneuver autopilot algorithm, under some states, once an executing state change event occurs, is then transformed into another executing state.This switching process can be expressed by a State Transferring equation:

δ:Q×E→Q

Wherein, Q={q 0, q 1..., q n,

E is all State Transferring events, at least should comprise run (execution), error (mistake), stop (stopping), restart (restarting), next_phase (next stage)

Maneuver autopilot algorithm is after selected, and can will begin in a minute execution.This time, maneuver autopilot algorithm was in incipient state.

Maneuver autopilot algorithm in the process of implementation, is transformed into next executing state by an executing state usually, until termination.

Each maneuver autopilot algorithm can end in two kinds of executing states:

● complete motor-driven final state: motor-driven target is reached.

● mistake final state: perform interruption for some reason.

As can be seen here, each executing state in a maneuver autopilot algorithm represents its different execute phase.Such as one maneuver autopilot algorithm of overtaking other vehicles can be divided into following state:

1. incipient state q0: wait for that passing track can pass through.

2.q1: enter passing track.

3.q2: arrive Overtaken Vehicle front certain distance.

4. complete motor-driven final state qf: original track is got back in lane change.

In the process performed, if the condition (state transition condition) entering next executing state meets, then enter next state.The example of state transition condition is:

● self-propelled vehicle reaches a specific position, and such as in the maneuver autopilot algorithm of overtaking other vehicles, motor vehicle motion has arrived passing track.

● the signal in surrounding environment there occurs change, and such as in the maneuver autopilot algorithm crossing crossing, traffic lights is green by red stain.

Meeting of state transition condition can allow certain error limit, and the condition such as having reached a specific position can allow actual position and target location to there is the deviation not affecting following maneuver autopilot algorithm and perform.

Meanwhile, each state includes the condition (state maintenance condition) maintaining this state and must meet.Whether maneuver autopilot algorithm constantly can detect these conditions in the process of implementation and meet.If these conditions are no longer satisfied, then maneuver autopilot algorithm can enter wrong final state.

The monitoring of state maintenance condition be the needs whether current environment of self-propelled vehicle meets current execution state Secure execution.Such as:

● condition that the forward visibility of sensor is blocked and cannot judges the vehicle condition of passing track under the q0 state of maneuver autopilot algorithm of overtaking other vehicles?

● condition that the speed of the vehicle come head-on is sailed in front and distance allows enough room and times to complete ensuing passing maneuver safely under the q1 state of maneuver autopilot algorithm of overtaking other vehicles, assuming that passing track is reversed carriage?

● under the q2 state of maneuver autopilot algorithm of overtaking other vehicles: the F-Zero in Overtaken Vehicle speed now and this section allows lane change to get back to original track?

Under some executing states, if maneuver autopilot algorithm determines to perform a specific action, then produce a performance-based objective instruction, input action control module, completed the output of search to path of motion and Mechanical course instruction by the latter.

Each maneuver autopilot algorithm, as overtaken other vehicles or crossing crossing, can parametrization (parameterize).It is the same that parametrization produces one group of behavior object, but the maneuver autopilot algorithm that driving style is different.Such as maneuver autopilot algorithm of overtaking other vehicles, can parameter turn to:

● closely, overtake other vehicles at a slow speed

● closely, overtake other vehicles fast

● remote, overtake other vehicles at a slow speed

● remote, overtake other vehicles fast

Cross crossing maneuver autopilot algorithm can parameter turn to:

● quick crossing excessively

● cross crossing at a slow speed

Arrive object point (travelling along road) maneuver autopilot algorithm can parameter turn to:

● arrive coordinate points fast

● arrive coordinate points at a slow speed

Follow front vehicles maneuver autopilot algorithm can parameter turn to:

● closely follow front vehicles

● follow front vehicles at a distance

As can be seen here, parameter determines the expected value/a reference value of some controlling quantitys in maneuver autopilot algorithm implementation.Such as maneuver autopilot algorithm of overtaking other vehicles:

● speed parameter of overtaking other vehicles determines the time taking passing track.

● when lane change is returned and the parameter of Overtaken Vehicle distance reflect the emergentness of lane change.

The current all information relevant with driving of real-time Driving Decision-making Algorithm Analysis, finds out a most suitable maneuver autopilot algorithm.This algorithm relates to above-mentioned 3 class data structures:

● virtual world event sets tuple.Motor vehicle driving decision-making and human brain decision process similar, the prerequisite of decision-making is the sufficient perception to surrounding environment.Virtual world event sets tuple represents the environmental information of the environmental information highly structural after treatment to perception.Here " environment " not only comprises the road conditions that on road, various impact is driven, also comprise the situation of self-propelled vehicle self (as tyre pressure sensor simultaneously, oil plant sensor, brake module sensor, the state of engine temperature sensor monitoring), also comprise the state of the various sensor equipment of self-propelled vehicle simultaneously, also comprise the state running on various software program on car-mounted computer simultaneously.

● route planning tuple, represents current total travel direction.

● maneuver autopilot algorithm set, represents the various maneuvers that Behavior-Based control module has the ability to perform.

Real-time Driving Decision-making algorithm is divided into two stages:

1. from the set of all maneuver autopilot algorithms, select the maneuver autopilot algorithm of safe and feasible.

2. in the maneuver autopilot algorithm of safe and feasible, select optimum maneuver autopilot algorithm.

The primary requirement of behaviour decision making module is exactly the behavior generated must safe and feasible.The condition that the maneuver autopilot algorithm of safe and feasible must meet simultaneously is:

● ensure the safety of other road users.

● ensure the safety of self-propelled vehicle self.

● observe traffic rules and regulations.

● to road conditions accuracy of judgement.

Different when, the maneuver autopilot algorithm meeting safe and feasible condition may have 0 to multiple.Below attempt several situation with the judgement of clear and definite safe and feasible condition.Assuming that in following several situation, the set of all maneuver autopilot algorithms is:

● arrive coordinate points (travelling along road)

● cross crossing

● overtake other vehicles

● follow front vehicles

● emergency braking

Situation 1:

Self-propelled vehicle travels along road, now also has suitable distance (such as, being greater than 250 meters) from crossing.Current lane front does not have other vehicles.

Now, the maneuver autopilot algorithm meeting safe and feasible condition should be:

● arrive coordinate points (travelling along road)

Other several maneuver autopilot algorithms do not meet because:

● cross crossing: current not near crossing.

● overtake other vehicles: front does not have vehicle to surpass.

● follow front vehicles: front does not have vehicle to follow.

● emergency braking: all is clear ahead.

Situation 2:

Self-propelled vehicle is followed after another car.

Now, the maneuver autopilot algorithm meeting safe and feasible condition should be:

● follow front vehicles.

● overtake other vehicles.

Other several maneuver autopilot algorithms do not meet because:

● cross crossing: current not near crossing.

● arrive coordinate points (travelling along road): the motion of self-propelled vehicle self is limited by front vehicles.

● emergency braking: preceding object thing is outside safety distance.

Situation 3:

Self-propelled vehicle is followed after another car, and lane mark is solid line.

Now, the maneuver autopilot algorithm meeting safe and feasible condition should be:

● follow front vehicles.

Maneuver autopilot algorithm of overtaking other vehicles no longer meets safe and feasible condition because passing track is unavailable.

Situation 4:

Self-propelled vehicle is followed after another car.The front speed of a motor vehicle 50 kilometers/hour.Lane mark is dotted line, speed limit 50 kilometers/hour.

Now, the maneuver autopilot algorithm meeting safe and feasible condition should be:

● follow front vehicles.

Maneuver autopilot algorithm of overtaking other vehicles no longer meets safe and feasible condition.Because if perform this maneuver autopilot algorithm, then front vehicles must be accelerated beyond, and front vehicles reaches the speed restriction of current road segment.

Situation 5:

A static obstacle (such as concrete wire pole) is within motor vehicle front safety distance:

Now, the maneuver autopilot algorithm uniquely meeting safe and feasible condition is emergency braking.

From situation above, this algorithm is it is of concern that maneuver autopilot algorithm whether safety and whether observing traffic rules and regulations, instead of the efficiency of maneuver autopilot algorithm and traveling comfort.The latter is screened by the optimum maneuver autopilot selection algorithm in next stage.

Therefore, the maneuver autopilot algorithm of safe and feasible is selected to need complete traffic regulation and motor vehicle driving knowledge.

While possessing above-mentioned knowledge, the maneuver autopilot algorithm of safe and feasible is selected also to need to make a policy in conjunction with current road conditions.

As can be seen here, in order to fully tackle in automatic Pilot the various situations that may encounter, select the maneuver autopilot algorithm of safe and feasible to need consideration a large amount of, interact extremely complicated factor.

The maneuver autopilot algorithm of safe and feasible is selected to use a Petri net (PetriNet) (as Fig. 3, hereinafter referred to as decision-making mode) to simulate so complicated decision process.

Petri net has strict mathematical expression mode, and the abundant math modeling for its process analysis procedure analysis.This makes intactly to check, analyze a Petri net becomes possibility, no matter this Petri net how complicated.So a very complicated process is extremely important for Driving Decision-making for this characteristic.

For the personal vehicle system of a self-propelled vehicle, safety is unanswerable primary index.And safety will be guaranteed, just must there is strict, reliable, complete test process.This test process wants can there be clear and definite, strict demonstration to the I/O/behavior of all possible case making policy decision nets.Mathematical expression form by decision-making mode relatively easily can realize these demonstrations, such as:

● under certain virtual world event condition, whether the storehouse of emergency braking maneuver autopilot algorithm can arrive (reachable).

● by analyzing Petri net boundedness (boundedness) characteristic, the number of the selected required virtual world event of certain maneuver autopilot algorithm can be learnt.

● can infer that decision-making mode can or can not produce self-contradictory decision-making.

Be illustrated in figure 4 a simple Petri net network, the element forming Petri net is comprise:

● storehouse institute (Place) circular node.

● transition (Transition) square nodes.

● directed arc (Arc) be storehouse and transition between directed arc.

● token (Token) be storehouse in dynamic object, another storehouse institute can be moved to from a storehouse.

The networking rule of Petri net element is:

● directed arc is directive.

● do not allow arc between two storehouse institutes or transition.

● storehouse can have the token of any amount.

● directed arc can link library institute and transition.

● each directed arc has a capability value, defaults to 1.

The state variation rule of Petri net is (as shown in Figure 5).

If when each input magazine institute (inputplace) of transition has quantity enough token, these transition are and are allowed to (enable).When transition are allowed to, transition will occur (fire), and the token of input magazine institute (inputplace) is consumed, and produce token for exporting storehouse institute (outputplace) simultaneously.

Simultaneously:

● transition be atom, that is, neither one transition only there occurs the possibility of half.

● there is the possibility that two or more transition are all allowed to, but transition once can only occur.In this case the order changing generation does not define.

● if there is transition, its input magazine number with export storehouse number unequal, the number of token will change, that is, token number nonconservation.

● Petri net is static, that is, there is not suddenly to emerge after there occurs transition another transition or storehouse institute, thus changes the possibility of Petri net structure.

● the state of Petri net by token in storehouse distribution determine.That is, just have the state determined time complete, the next transition of transition generation are waited for and being occurred, occurring when changing is the state that neither one is determined.

Be below the prototype structure of some Petri nets, as shown in Figure 6, may be used for the structure of decision-making mode.

Meanwhile, decision-making mode can use multi input, multi output structure as shown in Figure 7.This structure is generally used for representing the causal relationship of multiple factor and multiple different result.

Decision-making mode can also use single input as shown in Figure 8, single output, multistage transition structure.(transitive) causal relationship can passed between the multiple factor of this structure general proxy.

Decision-making mode is structurally divided into two subnets:

● traffic regulation and driving knowledge subnet (hereinafter referred to as subnet 1).

● route planning subnet (hereinafter referred to as subnet 2).

Shown in Fig. 9 is the structure (only the structure thinking of exemplary decision net) that the self-propelled vehicle Decision Control method of a decision-making mode local is possible:

Therefore, the implementation of decision-making mode is divided into two steps:

● the input data of the first step are the virtual world event sets tuple that virtual world event generator generates.After subnet 1 processes, obtain the safe and feasible maneuver autopilot algorithm meeting traffic regulation.

● on the basis of the maneuver autopilot algorithm that second step filters out in the first step, further consider route planning.The two through the process of subnet 2, draws safe and feasible and maneuver autopilot algorithm reconcilable with route planning as input data.

Before decision-making mode performs each time, the state according to virtual world event sets tuple and route planning tuple arranges token in the input magazine institute corresponding to its element.The corresponding input magazine institute of element of each virtual world event tuple and route planning tuple.If the state of this element is true, then corresponding input magazine is endowed a token.

After executing subnet 1, middle driving maneuver algorithms library gather in all feasible maneuver autopilot algorithms all can have a token.

Then, after executing subnet 2 further, export driving maneuver algorithms library gather in have the storehouse of token to do the maneuver autopilot algorithm of the safe and feasible finally determined.

If after executing whole decision-making mode, selected without any maneuver autopilot algorithm, then motive vehicle meeting conservatively stops, and waits for the decision-making of a new Event triggered new round, until a certain take turns decision-making after have motor-driven control algorithm selected.

If only have a maneuver autopilot algorithm selected, then this maneuver autopilot algorithm is directly executed.Otherwise if multiple maneuver autopilot algorithm is selected, behaviour decision making module enters the next stage, select from these maneuver autopilot algorithms one optimum.

Real-time Driving Decision-making algorithm is triggered and is called a performance period each time.The condition triggering the performance period can be:

● the maneuver autopilot algorithm of current execution arrives a final state (completing motor-driven final state or wrong final state).

● virtual world event sets tuple there occurs change (such as wherein the state of some virtual world event changes).

Once a selected execution of maneuver autopilot algorithm, as long as then still selected in the new performance period, so this maneuver autopilot algorithm will continue execution and goes down, until termination.

After each decision-making completes, all tokens are eliminated, for decision-making is prepared next time.

Any one maneuver autopilot algorithm all may terminate with error condition.Now, decision-making module can proceed to execution error recovery maneuver autopilot algorithm.Fault recovery maneuver autopilot algorithm is structurally consistent with other maneuver autopilot algorithm.Its object is under mainly the normal state of safety got back to by controlling machine motor-car from certain unexpected state.Maneuver autopilot algorithm enters wrong final state and is usually triggered by some concrete condition.If these conditions determine a specific Fault recovery maneuver autopilot algorithm very clearly, then behaviour decision making module directly performs this Fault recovery maneuver autopilot algorithm.Otherwise behaviour decision making module starts the performance period of a new round, according to new status information, select the maneuver autopilot algorithm that new.

Especially, optimum maneuver autopilot selection algorithm is self-propelled vehicle Decision Control method many criteria decision algorithm (MultipleCriteriaDecisionMaking), its objective is in a series of alternative maneuver autopilot algorithm, according to certain condition, select optimum maneuver autopilot algorithm.

It should be noted that, in this article, the such as relational terms of first and second grades and so on is only used for an entity or operation to separate with another entity or operational zone, and not necessarily requires or imply the relation that there is any this reality between these entities or operation or sequentially.And, term " comprises ", " comprising " or its any other variant are intended to contain comprising of nonexcludability, thus make to comprise the process of a series of key element, method, article or final terminal and not only comprise those key elements, but also comprise other key elements clearly do not listed, or also comprise by the intrinsic key element of this process, method, article or final terminal.When not more restrictions, the key element limited by statement " comprising ... " or " comprising ... ", and be not precluded within process, method, article or the final terminal comprising described key element and also there is other key element.In addition, in this article, " be greater than ", " being less than ", " exceeding " etc. be interpreted as and do not comprise this number; " more than ", " below ", " within " etc. be interpreted as and comprise this number.

Those skilled in the art should understand, the various embodiments described above can be provided as method, device or computer program.These embodiments can adopt the form of complete hardware embodiment, completely software implementation or the embodiment in conjunction with software and hardware aspect.The hardware that all or part of step in the method that the various embodiments described above relate to can carry out instruction relevant by program has come, described program can be stored in the storage medium that computer equipment can read, for performing all or part of step described in the various embodiments described above method.Described computer equipment, includes but not limited to: Personal Computer, server, general computer, single-purpose computer, the network equipment, embedded device, programmable device, intelligent mobile terminal, intelligent home device, wearable intelligent equipment, vehicle intelligent equipment etc.; Described storage medium, includes but not limited to: the storage of RAM, ROM, magnetic disc, tape, CD, flash memory, USB flash disk, portable hard drive, storage card, memory stick, the webserver, network cloud storage etc.

The various embodiments described above describe with reference to the diagram of circuit of method, equipment (system) and computer program according to embodiment and/or block scheme.Should understand can by the combination of the flow process in each flow process in computer program instructions realization flow figure and/or block scheme and/or square frame and diagram of circuit and/or block scheme and/or square frame.These computer program instructions can being provided to the treater of computer equipment to produce a machine, making the instruction performed by the treater of computer equipment produce device for realizing the function of specifying in diagram of circuit flow process or multiple flow process and/or block scheme square frame or multiple square frame.

These computer program instructions also can be stored in can in the computer equipment readable memory that works in a specific way of vectoring computer equipment, the instruction making to be stored in this computer equipment readable memory produces the manufacture comprising command device, and this command device realizes the function of specifying in diagram of circuit flow process or multiple flow process and/or block scheme square frame or multiple square frame.

These computer program instructions also can be loaded on computer equipment, make to perform sequence of operations step on a computing device to produce computer implemented process, thus the instruction performed on a computing device is provided for the step realizing the function of specifying in diagram of circuit flow process or multiple flow process and/or block scheme square frame or multiple square frame.

Although be described the various embodiments described above; but those skilled in the art are once obtain the basic creative concept of cicada; then can make other change and amendment to these embodiments; so the foregoing is only embodiments of the invention; not thereby scope of patent protection of the present invention is limited; every utilize specification sheets of the present invention and accompanying drawing content to do equivalent structure or equivalent flow process conversion; or be directly or indirectly used in other relevant technical fields, be all in like manner included within scope of patent protection of the present invention.

Claims (8)

1. self-propelled vehicle Decision Control method, is applied to the behaviour decision making module of self-propelled vehicle, it is characterized in that, comprise the steps:
Obtain the event information of virtual world module input;
Obtain the routing information of navigation module input;
According to event information and routing information generation machine motor-car maneuver,
When the road conditions of the information on electronic chart and reality are inconsistent, export traffic information to navigation module.
2. self-propelled vehicle Decision Control method according to claim 1, is characterized in that, also comprise the steps:
The shape of road of reappraising and trend;
Export road information to guidance module.
3. self-propelled vehicle Decision Control method according to claim 1, it is characterized in that, described event information comprises virtual world event, and described virtual world event comprises:
The virtual world event of obstacle class, the virtual world event of vehicle oneself state, the virtual world event of vehicle current driving road segment or the virtual world event of perception aspect.
4. self-propelled vehicle Decision Control method according to claim 1, is characterized in that,
Analyze the routing information of navigation module input, therefrom extract the route planning of motor-driven vehicle going;
Wherein, route planning comprise keep straight on forward, turn left forward, turn right forward or turn around travel.
5. self-propelled vehicle Decision Control method according to claim 1, is characterized in that, described " generation machine motor-car maneuver " comprises according to finite state automaton generation machine motor-car maneuver.
6. self-propelled vehicle Decision Control method according to claim 1, is characterized in that, described " generation machine motor-car maneuver " comprises the self-propelled vehicle maneuver that use Petri net generates safety.
7. self-propelled vehicle Decision Control method according to claim 1, is characterized in that, described " generation machine motor-car maneuver " comprises the steps:
According to event information and traffic regulation and drive the maneuver autopilot method that knowledge information generates safety;
Safe and feasible is generated and maneuver autopilot method reconcilable with route planning according to the maneuver autopilot method generated and route planning subnet information.
8. self-propelled vehicle Decision Control method according to claim 1, is characterized in that, also comprise step:
When receiving error condition;
Judge whether the Fault recovery maneuver autopilot algorithm meeting this error condition, have, perform this Fault recovery maneuver autopilot algorithm;
Otherwise, start the performance period of a new round, according to new status information, select the maneuver autopilot algorithm that is new.
CN201510081698.9A 2015-02-15 2015-02-15 Self-propelled vehicle Decision Control method CN104760592B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201510081698.9A CN104760592B (en) 2015-02-15 2015-02-15 Self-propelled vehicle Decision Control method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201510081698.9A CN104760592B (en) 2015-02-15 2015-02-15 Self-propelled vehicle Decision Control method

Publications (2)

Publication Number Publication Date
CN104760592A CN104760592A (en) 2015-07-08
CN104760592B true CN104760592B (en) 2016-02-17

Family

ID=53642784

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201510081698.9A CN104760592B (en) 2015-02-15 2015-02-15 Self-propelled vehicle Decision Control method

Country Status (1)

Country Link
CN (1) CN104760592B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2019037125A1 (en) * 2017-08-25 2019-02-28 深圳市得道健康管理有限公司 Artificial intelligence terminal and behavior control method thereof

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
DE102005055744A1 (en) * 2005-11-23 2007-05-24 Robert Bosch Gmbh Driver assisting system for motor vehicle, has comparison module generating information for guiding vehicle based on dynamic characteristics measured by dynamic module, where information is displayed by display device
CN202011393U (en) * 2011-03-05 2011-10-19 山东申普交通科技有限公司 Full-automatic manned guide vehicle based on image recognition technology
CN103935361A (en) * 2013-01-21 2014-07-23 通用汽车环球科技运作有限责任公司 Efficient data flow algorithms for autonomous lane changing, passing and overtaking behaviors

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR101194603B1 (en) * 2009-03-12 2012-10-25 한국전자통신연구원 Unmanned transport apparatus and its method

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
DE102005055744A1 (en) * 2005-11-23 2007-05-24 Robert Bosch Gmbh Driver assisting system for motor vehicle, has comparison module generating information for guiding vehicle based on dynamic characteristics measured by dynamic module, where information is displayed by display device
CN202011393U (en) * 2011-03-05 2011-10-19 山东申普交通科技有限公司 Full-automatic manned guide vehicle based on image recognition technology
CN103935361A (en) * 2013-01-21 2014-07-23 通用汽车环球科技运作有限责任公司 Efficient data flow algorithms for autonomous lane changing, passing and overtaking behaviors

Also Published As

Publication number Publication date
CN104760592A (en) 2015-07-08

Similar Documents

Publication Publication Date Title
US10379533B2 (en) System and method for autonomous vehicle fleet routing
CN106251016B (en) A kind of parking system paths planning method based on dynamic time windows
CN103158705B (en) Method and system for controlling a host vehicle
Guanetti et al. Control of connected and automated vehicles: State of the art and future challenges
US20200004241A1 (en) Teleoperation system and method for trajectory modification of autonomous vehicles
González et al. A review of motion planning techniques for automated vehicles
You et al. Trajectory planning and tracking control for autonomous lane change maneuver based on the cooperative vehicle infrastructure system
CN106114507B (en) Local path planning method and device for intelligent vehicle
Campbell et al. Autonomous driving in urban environments: approaches, lessons and challenges
Loos et al. Adaptive cruise control: Hybrid, distributed, and now formally verified
Richards et al. Aircraft trajectory planning with collision avoidance using mixed integer linear programming
WO2017079219A1 (en) Teleoperation system and method for trajectory modification of autonomous vehicles
Huang et al. Assessing the mobility and environmental benefits of reservation-based intelligent intersections using an integrated simulator
Varaiya et al. Sketch of an IVHS systems architecture
Urmson et al. Autonomous driving in traffic: Boss and the urban challenge
Tomlin et al. A game theoretic approach to controller design for hybrid systems
CN108698595A (en) The control system of method and vehicle for controlling vehicle movement
Gelenbe et al. Simulation with learning agents
CN108292134A (en) Machine learning system and technology for optimizing remote operation and/or planner decision
Digani et al. Ensemble coordination approach in multi-agv systems applied to industrial warehouses
Ehlert et al. Microscopic traffic simulation with reactive driving agents
Faust et al. PRM-RL: Long-range robotic navigation tasks by combining reinforcement learning and sampling-based planning
CN205582269U (en) Intelligent traffic control system
CN104574953A (en) Traffic signal prediction
CN108829087B (en) A kind of intelligent test system of autonomous driving vehicle

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
EXSB Decision made by sipo to initiate substantive examination
SE01 Entry into force of request for substantive examination
C14 Grant of patent or utility model
GR01 Patent grant