TWI583581B - Automatic Driving System with Driving Behavior Decision and Its - Google Patents

Description

Autopilot system with driving behavior decision and method thereof

The present invention relates to a related art of automatic driving, and particularly relates to an automatic driving system and method for determining driving behaviors for determining the best obstacle avoidance behavior.

In order to find a safer driving style and a more efficient road use environment, major automakers have been actively investing in the development of automatic assisted driving or automatic driving systems so that the vehicle can automatically assist the driver in making decisions or intervening to control the vehicle. Starting from a preventive perspective, you can achieve the goal of avoiding accidents directly.

The general automatic driving system uses the sensor to obtain the surrounding environment, assists in driving the vehicle or directly controls the vehicle to avoid the collision avoidance behavior, and can effectively reduce the collision risk. However, the current automatic driving decision method is as follows. If there is an object in front, the general automatic driving decision logic can only use the following two points to judge. 1. When the available space is found, the automatic driving system can advance toward the travelable space. 2. No driving space is found, and the automatic driving system generates an error message indicating that there is no driving space ahead, indicating that it is not possible to move forward. However, since it is judged that the security is too arbitrary based on the data of the travelable space, and the judgment basis is too narrow, it is more likely to cause an increase in the computational complexity of the subsequent path calculation module, so it is quite insecure to use this judgment method.

In view of the above, the present invention provides an automatic driving system with a driving behavior decision-making function and a method thereof for effectively avoiding the above-mentioned problems. These questions.

The main object of the present invention is to provide an automatic driving system with driving behavior decision-making and a method thereof, which use various judgment methods to improve the safety of driving behavior, reduce the complexity of path calculation, and generate safer mobile behavior. For the vehicle to move forward.

Another object of the present invention is to provide an automatic driving system with a driving behavior decision and a method thereof, which are to vectorize all objects detected by a road to calculate an object safety determination, and give a safety weight according to the risk of collision, and further Infer the safety of road space, give space security weights, and make driving decisions such as left turn, right turn, forward, and brake.

In order to achieve the above object, the present invention provides a method for determining driving behavior of an automatic driving. The method comprises the steps of: first generating a left turn signal, a straight line signal and a right turn signal through a processor, and collecting one of the vehicle bodies through a detecting device. The body movement signal and the object movement signal of the plurality of objects; the processor converts the body movement signal and the plurality of object movement signals into a body movement vector and a plurality of object movement vectors respectively; the processor determines each object according to the body movement vector and the plurality of object movement vectors. Whether it is a dangerous object, if it is judged that the object is a dangerous object, the sum of the collision time of all the objects and the vehicle body, and the ratio of the collision time between the dangerous object and the vehicle body, generates a weight of the dangerous object, and if not, judges that the object is a non-hazardous object. , using the distance between the non-hazardous object and the vehicle body to generate a non-hazardous object weight; the processor then defines the left turn road area of the left turn signal, the straight road area of the straight signal and the right turn road area of the right turn signal, and according to Dangerous objects passing through the road area The weights of the weighted and non-hazardous objects are respectively used to determine the weights of the left-turn road area, the straight road area, and the right-turn road area; the processor generates a left-turn signal weight by using the weights of the left-turn road area, the straight road area, and the right-turn road area, Always signal weight and a right-turn signal weight; the processor takes the left-turn signal weight, the straight-line signal weight, and the highest weight of the right-turn signal weight to determine the highest weight signal. Whether the weight of the number is greater than a preset value of the weight, and if so, a forward signal is generated according to the direction of the highest weight signal; if not, a traffic signal is generated.

In addition, the present invention also provides an automatic driving system with a driving behavior decision and a method thereof, including a processor electrically connected to an object detecting device, a body moving and picking device, and a storage device, wherein the processor can generate a left turn Signal, continuous signal and a right-turn signal, and capture the body motion signal generated by the body movement pick-up device and the plurality of object motion signals generated by the object detection device, respectively converted into a body motion vector and a plurality of object motion vectors The processor then determines whether the object is a dangerous object according to the vehicle movement vector and the plurality of object movement vectors. If it is a dangerous object, the weight object equation of the dangerous object in the storage device is extracted, and the weight of the dangerous object is calculated, and if it is a non-hazardous object, the object is captured. The weighting equation of the non-hazardous object in the storage device calculates a weight of the non-hazardous object; the processor can calculate the spatial weighting equation in the storage device, and substitute the weight of the dangerous object or the weight of the non-hazardous object into the spatial weight equation to calculate the left turn separately. The left turn of the signal, the straight road of the straight signal The weight of the area and the right turn road area of the right turn signal, the processor draws the signal weight equation, and assigns the weights of the left turn road area, the straight road area, and the right turn road area into the signal weight equation to generate a left turn signal weight, The signal weight and the right turn signal weight are always used. The processor takes the signal with the highest weight in the left turn road area, the straight road area and the right turn road area. If the weight of the highest weight signal is greater than a weight preset value, then A forward signal is generated according to the direction of the highest weight signal to advance a vehicle. If the weight of the highest weight signal is less than a weight preset value, a vehicle signal is generated to stop the vehicle from advancing.

The purpose, technical content, features and effects achieved by the present invention will be more readily understood by the detailed description of the embodiments.

1‧‧‧Automatic driving system with driving behavior decision

10‧‧‧ Object detection device

12‧‧‧ Body moving device

14‧‧‧Storage device

16‧‧‧ Processor

18‧‧‧ objects

18’‧‧‧ objects

18"‧‧‧ objects

20‧‧‧Self vehicle

32‧‧‧Left to road area

32’‧‧‧ Turn left road area

34‧‧‧ Straight road area

36‧‧‧Right turn road area

A‧‧‧left turn signal

b‧‧‧Direct signal

C‧‧‧right turn signal

The first figure is a block diagram of an automatic driving system in accordance with an embodiment of the present invention.

The second figure is a flowchart of a path decision method according to an embodiment of the present invention.

The third figure and the embodiment of the present invention generate a pre-selected forward signal schematic.

The fourth figure is a schematic diagram of the decision forward signal of the embodiment of the present invention.

Referring to the first figure, the automatic assist driving system 1 of the present embodiment can be disposed on a self-vehicle, wherein the automatic driving system 1 with driving behavior decision includes a detecting device, and the detecting device in the embodiment is an object detecting device. 10 and a body moving and picking device 12, the object detecting device 10 is to take a moving signal of an object other than the own vehicle to generate a plurality of object moving signals; and the body moving and picking device 12 is to take the moving signal of the own vehicle. To generate a body movement signal; a storage device 14 stores a dangerous object determination equation, a dangerous object weight equation, a non-hazardous object weight equation, a spatial weight equation, and a signal weight equation; a processor 16 is electrically connected to the above The object detecting device 10, the vehicle body moving and picking device 12, and the storage device 14 are configured to receive a plurality of object moving signals, a body moving signal, and a dangerous object determining equation, a dangerous object weighting equation, and a non-hazardous object in the storage device 14. Weight equations, spatial weight equations, and signal weight equations, By vectorizing the object motion signal, it is brought into the dangerous object judgment equation, and a travel path is generated according to the demanded dangerous object weight equation, the non-hazardous object weight equation, the spatial weight equation, and the signal weight equation for the vehicle to follow. The best way forward.

Next, please refer to the first to third figures to illustrate how to use the above-described automatic driving system 1 for driving behavior decision to judge to generate a safe driving behavior to assist the vehicle to advance according to a safe path. First, in step S10, the processor 16 generates a left turn signal a, a continuous signal b, and a right turn signal c, and detects the plurality of objects 18, 18', 18" around the own vehicle 20 through the object detecting device 10. The object moves the signal, and detects a body movement signal of the own vehicle 20 through the vehicle body moving and capturing device 12. Next, in step S12, the processor 16 converts the body movement signal when receiving the body movement signal and the plurality of object movement signals. For the vehicle body motion vector, and converting the plurality of object motion signals into a plurality of object motion vectors. Then, in step S14, the processor 16 retrieves the dangerous object judgment equations in the storage device 14, and determines whether each object 18, 18', 18" is one by one. For dangerous objects, the judgment method is to substitute the vehicle movement vector and the complex object motion vector into the dangerous object judgment equation. The dangerous object judgment equation (1) is as follows: among them Is the body movement vector; It is the object movement vector; the δ system is a distance preset value. among them ≠0 is to determine whether the own vehicle 20 is parallel to the objects 18, 18', 18", if it is parallel, it means 0, if it is not parallel, it is not equal to 0; <0) determines whether the own vehicle 20 and the objects 18, 18', 18" are in the same direction or in the reverse direction. If it is greater than 0, it means that the own vehicle 20 and the objects 18, 18', 18" are advanced in the same direction, if less than 0 means that the own vehicle 20 and the object system 18, 18', 18" advance in the reverse direction, and may collide; | ) < δ determines whether the distance between the own vehicle 20 and the objects 18, 18', 18" is less than a distance preset value, and if it is less than the preset value, the distance between the own vehicle 20 and the objects 18, 18', 18" Too close to there is the possibility of a collision. So when [ ≠0] indicates that the own vehicle 20 is not parallel to the objects 18, 18', 18", or [( =0)∩ ( <0)] Although the own vehicle 20 is parallel to the objects 18, 18', 18", the own vehicle 20 and the objects 18, 18', 18" are advanced in a reverse manner, or [( =0)∩(|( )|< δ )] although the own vehicle 20 is parallel to the objects 18, 18', 18" but the distance between the own vehicle 20 and the objects 18, 18', 18" is less than a predetermined distance, if it is a front object, The distance preset value can be set according to the vehicle distance. If it is the left or right object, the distance preset value can be set by one lane distance. As long as one of the above conditions is true, the object 18, 18', 18" is indicated. For a dangerous object, the present embodiment determines that the dangerous object is the object 18', 18"; after determining the dangerous object 18', 18", the process proceeds to step S16, and the processor 16 calculates the dangerous object weight equation in the storage device 14 to calculate the dangerous object. 18', 18" weights, dangerous object weight equation (2) is as follows: Where W _U is the weight of the dangerous object 18' or 18"; C _U is the collision time of the dangerous object 18' or 18" with the own vehicle; C _t is the collision of the complex object 18, 18', 18" with the own vehicle 20 The sum of the time. The processor 16 utilizes the sum of the collision times of all objects 18, 18', 18" with the own vehicle 20, and the collision time of the dangerous object 18' with the own vehicle 20 or the collision of the dangerous object 18" with the own vehicle 20. The ratio of time produces the weight of the dangerous objects 18', 18".

However, if the processor 16 determines that the condition of the dangerous object determination equation (1) is not satisfied at step S14, it indicates that the object 18 is a non-hazardous object. In this embodiment, the object 18 is a non-hazardous object, and the object 18 is determined to be In the case of a non-hazardous object, the process proceeds to step S18, and the non-hazardous object weight of the non-hazardous object 18 is calculated. The processor 16 extracts the non-hazardous object weighting equation in the storage device 14 to generate a non-dangerous object 18 from the own vehicle 20. The weight of dangerous objects, where the weight equation of non-hazardous objects (3) is expressed as: Where W _N is the weight of the non-hazardous object; d is the distance between the non-hazardous object and the vehicle body, μ is the constant generated by d, and the μ value is proportional to d. In general, the value of the weight of a non-hazardous object will be greater than the value of the weight of the dangerous object.

Next, as shown in step S20, and with reference to the fourth figure, the dangerous object weights of the dangerous objects 18', 18" are calculated through step S16, and the non-hazardous object weights of the non-hazardous objects 18 are calculated in step S18, and then processed. The processor 16 defines a plurality of left turn road areas 32 according to the left turn signal a, a plurality of straight road areas 34 defined according to the straight line signal b, and a plurality of right turn road areas 36 defined according to the right turn signal c. The processor 16 retrieves the storage device 14 The space weight equation in the middle determines the weight of each of the left turn road area 32, the straight road area 34, and the right turn road area 36, wherein the spatial weight equation (4) is expressed as: The weight of the W _R road area, W _O is the weight of the dangerous object or the weight of the non-hazardous object. This embodiment includes the weight of the dangerous object and the weight of the non-hazardous object; D _O is the distance between the center point of the road area and the object; φ is one The constant is proportional to D _O . The weight of each road area is respectively taken from the left turn road area 32, the straight road area 34, and the right turn road area 36 via the processor 16, and the objects 18, 18', 18 that are closest to the road area to pass through the road area. The weight of the "plus" plus a value proportional to the distance relationship between the center point of each of the left-turn road area 32, the straight road area 34, and the right-turn road area 36 and the objects 18, 18', 18". Taking the plural left turn road area 32 as an example, the object system closest to the center point among the objects of one of the left turn road areas 32' is first calculated as the non-hazardous object 18, so the non-hazardous object weights of the non-hazardous objects 18 are taken. And adding the constant φ of the left turn road area 32' and the object 18, calculate the weight of the left turn road area 32', and the calculation of the next left turn road area 32 is the same, so the description is not repeated, of course, if you want to calculate When the straight road area 34 is taken, the weight of the dangerous object 18' is taken, and the constant point φ of the center point of the straight road area 34 and the object 18' is added, and the weight of the straight road area 34 is calculated, and the next straight road is calculated. The calculation of the area 34 is the same, so the retelling is not repeated. The weight of the right turn road area is the weight of the dangerous object 18", and the constant φ of the right turn road area 36 and the object 18" is calculated. The weight of the turning road area 36, and the calculation of the next right turn road area 36 are the same, so the description is not repeated.

Next, proceeding to step S22, the processor 16 retrieves the signal weighting equation in the storage device 14, and passes each of the left-turning road regions 32 according to the left-turning signal a, and the straight-line signal b passes through each of the straight road regions 34 and turns right. The signal c passes through the weight of each right turn road area 36, and generates a left turn signal weight, a straight line signal weight and a right turn signal weight respectively. The signal weight equation (5) is as follows: B _i =min W _R (5) Where W _R is the road area weight, and B _i is the signal weight. In this embodiment, the left-turn signal weight, the straight-line signal weight, and the right-turn signal weight are included in the example, and the judgment manner is based on the left-turn signal a to the left. The turn road area 32, the straight line signal b in the straight road area 34, and the right turn signal c in the right turn road area 36 have the minimum road area weight as the signal weight calculation, taking the plurality of left turn road areas 32, 32' as an example. Since the object 18 is a straight forward object 18, if the own vehicle 20 wants to advance according to the left turn signal a, when traveling to the left turn road area 32', it may collide with the object 18, so the left turn road area 32' is judged to be the most unsafe point in the plurality of left turn road areas 32, so the least secure left turn road area 32' is used as the left turn signal weight, because its behavior is safe to observe whether the entire path is safe, so The overall behavioral safety is represented by the most dangerous road area 32'.

Next, proceeding to step S24, comparing the left-turn signal weight, the straight-line signal weight, and the right-turn signal weight to obtain the signal with the highest weight. In this embodiment, the left-turn signal a is the signal with the highest weight, and the processor 16 determines the left turn. Whether the signal weight is greater than a weight preset value, if yes, proceeding to step S26 to generate a forward signal according to the left turn road signal direction, so that the own vehicle 20 advances according to the left turn signal a; if the left turn signal weight is less than a weight preset value, Then, a brake signal is generated as shown in step S28 for a self-vehicle 20 to stop.

In summary, the present invention can use a variety of judgment methods, which can generate safer mobile behavior for the vehicle to follow the advancement, and the judgment method is to vectorize all the objects detected by the road to calculate the object safety judgment according to the collision risk. The safety weight is given to the high and low, and the safety of the road space is calculated, and the space security weight is given for the driving behaviors such as left turn, right turn, forward travel, and braking, which can effectively improve the safety of driving, and at the same time, the complexity of low path calculation. degree.

The above is only the preferred embodiment of the present invention, and is not intended to be limiting. The scope of the practice of the invention is defined. Therefore, any changes or modifications of the features and spirits of the present invention should be included in the scope of the present invention.

1‧‧‧Automatic driving system with driving behavior decision

10‧‧‧ Object detection device

12‧‧‧ Body moving device

14‧‧‧Storage device

16‧‧‧ Processor

Claims (10)

An autopilot driving behavior decision method includes the steps of: generating a left turn signal, a continuous signal, and a right turn signal through a processor, and collecting a body motion signal of a vehicle body and a plurality of objects of a plurality of objects through a detecting device a mobile signal; the body motion signal and the object motion signals are respectively converted into a body motion vector and a plurality of object motion vectors by the processor; the processor determines each object according to the body motion vector and the object motion vector Whether it is a dangerous object or a non-hazardous object, and generates a dangerous object weight or a non-hazardous object weight according to the dangerous object or the non-hazardous object; the processor respectively defines the left turn road area of the left turn signal, and the straight line of the straight line signal a road area and a right turn road area of the right turn signal, and determining a weight of the left turn road area, the straight road area, and the right turn road area according to the dangerous object weight or the non-hazardous object weight to be passed through the road area; And generate a left turn signal weight, always signal right And a right-turning signal weight for determining the highest weight signal, determining whether the weight of the highest weight signal is greater than a weight preset value: if yes, generating a forward signal according to the highest weight signal direction; and No, a car signal is generated. The method for determining driving behavior of an automatic driving according to claim 1, wherein determining whether the object is a dangerous object determines whether the object is a dangerous object through a dangerous object judgment equation, and the dangerous object determining equation is expressed as: Which should Is the body movement vector; The vector is moved by the object; the δ is a distance preset value. The method for determining driving behavior of an automatic driving according to claim 1, wherein generating a dangerous object weight or a non-hazardous object weight is based on determining whether the object is the dangerous object, and using all of the objects to collide with the vehicle body. The sum of the sum of the dangerous objects and the collision time of the vehicle body generates the weight of the dangerous object, which is generated by a dangerous object weighting equation, which is expressed as: Wherein the W _U is the weight of the dangerous object; the C _U is the collision time of the dangerous object with the vehicle body; the C _t is the sum of the collision times of the objects with the vehicle body; if the object is a non-hazardous object And using the distance between the non-hazardous object and the vehicle body to generate the non-hazardous object weight, which is generated by a non-hazardous object weighting equation, and the non-hazardous object weight is expressed as: Wherein the W _N is the non-hazardous object weight; the d is the distance between the non-hazardous object and the vehicle body, and the μ is a constant generated by d, and the μ value is proportional to d. The driving behavior decision method of the automatic driving according to claim 1, wherein the weight of the left turn road area, the straight road area, and the right turn road area is determined according to the dangerous object weight or the non-hazardous object weight by a spatial weight The equation judges that the spatial weight equation is expressed as: Wherein the W _R is a weight of the road area, the _WO is the object weight; the D _O is the distance between the center point of the road area and the object; and the φ is a constant proportional to D _O . The method for determining driving behavior of an automatic driving according to claim 1, wherein the left turn signal weight value, the straight line signal weight value, and the right turn signal weight are generated, the left turn road area, the straight road area, and the The weight of the right turn road area is substituted into a signal weighting equation to respectively generate the left turn signal weight, the straight line signal weight and the right turn signal weight, and the signal weight equation is expressed as: B _i =min W _R ; wherein the B _i It is the weight of the signal, and the W _R is the weight of the road area. An automatic driving system with driving behavior decision-making includes: an object detecting device for generating a plurality of object moving signals; a body moving picking device for generating a body moving signal; and a storage device for storing a dangerous object determining equation and a danger An object weighting equation, a non-hazardous object weighting equation, and a spatial weighting equation; and a processor electrically connected to the object detecting device, the vehicle body moving and extracting device, and the storage device, the processor generating a left turn signal, a signal and a right turn signal, and the body motion signal and the motion signals of the objects are respectively converted into a body motion vector and a plurality of object motion vectors, and each body motion vector and the object motion vector are used to determine each Whether the object is a dangerous object or a non-hazardous object, and calculates a weight of a dangerous object or a weight of a non-hazardous object; the processor extracts the spatial weighting equation, and substitutes the weight of the dangerous object or the weight of the non-hazardous object into the spatial weight Equation to calculate the left turn of the left turn signal Regional road, straight road area of the straight signal and the right The weight of the right-turning road area of the transfer number, and generates a left-turn signal weight, a continuous signal weight, and a right-turn signal weight, the processor capturing the left-turn road area, the straight road area, and the right-turn road area The highest weight signal, if the weight of the highest weight signal is greater than a weight preset value, a forward signal is generated according to the direction of the highest weight signal to advance a vehicle if the weight of the highest weight signal is less than one The weight preset value generates a traffic signal to stop the vehicle from moving forward. The automatic driving system with driving behavior decision according to claim 6, wherein the determining whether the object is a dangerous object determines whether the object is a dangerous object through a dangerous object judgment equation, and the dangerous object determining equation is expressed as: Which should Is the body movement vector; The vector is moved by the object; the δ is a distance preset value. An automatic driving system having a driving behavior decision according to claim 6, wherein generating a dangerous object weight or a non-hazardous object weight is based on determining that the object is the dangerous object, and using all of the objects to collide with the vehicle body The sum of time, and the ratio of the collision time of the dangerous object to the vehicle body, generates the weight of the dangerous object, which is generated by a dangerous object weighting equation, which is expressed as: Wherein the W _U is the weight of the dangerous object; the C _U is the collision time of the dangerous object with the vehicle body; the C _t is the sum of the collision times of the objects with the vehicle body; if the object is a non-hazardous object And using the distance between the non-hazardous object and the vehicle body to generate the non-hazardous object weight of the non-hazardous object weight, which is generated by a non-hazardous object weighting equation, the non-hazardous object weight is expressed as: Wherein the W _N is the non-hazardous object weight; the d is the distance between the non-hazardous object and the vehicle body; the μ is a constant generated by d, and the μ value is proportional to d. The automatic driving system with driving behavior decision according to claim 6, wherein the spatial weight equation is expressed according to the spatial weight equation: Wherein the W _{R is} the road area weight, the _WO is the object weight; the D _O is the distance between the center point of the road area and the object; the φ is a constant proportional to D _O . An automatic driving system with a driving behavior decision according to claim 6, wherein the storage device further stores a signal weighting equation, the processor extracts the signal weighting equation, and the left turn road area, the straight road area, and The weight of the right turn road area is substituted into the signal weighting equation to generate the left turn signal weight, the straight line signal weight and the right turn signal weight respectively, and the signal weight equation is expressed as: B _i =min W _R ; wherein the B _i is the signal weight, and the W _R is the road area weight.