Obstacle recognition method, device and electronic equipment

Abstract

The present invention provides an obstacle recognition method, device and electronic device, which first obtains the attribute information of the target obstacle on the current road; the attribute information includes at least one of the value attribute of the target obstacle, the damage degree of the target obstacle when the target obstacle contacts the current vehicle, and the damage degree of the current vehicle when the target obstacle contacts the current vehicle; then, based on the attribute information, the valuation parameter of the target obstacle is determined; finally, based on the valuation parameter, the impact category of the target obstacle is determined; wherein the impact category is used to indicate the impact degree of the target obstacle on the driving route of the current vehicle. This method estimates the obstacle through the attribute information of the obstacle, and then determines the impact degree of the obstacle on the driving route of the current vehicle, accurately determines the impact degree of the road obstacle on the driving route of the vehicle, and thus improves the driving efficiency and safety of the vehicle.

Description

Technical Field

[0001] The present invention relates to the field of autonomous driving technology, and in particular to an obstacle recognition method, device and electronic equipment. Background Art

[0002] In related technologies, when an autonomous driving vehicle is driving on the road, it usually regards all obstacles detected on the road as non-collisionable and non-crushable, and thus plans the driving route to avoid all obstacles. This method may cause sudden braking or sharp turns on the driving route, and the driving route is relatively complicated, resulting in low driving efficiency of the vehicle and affecting the riding experience of passengers. Summary of the invention

[0003] In view of this, an object of the present invention is to provide an obstacle recognition method, device and electronic equipment to accurately determine the impact of road obstacles on the vehicle's driving route, thereby improving the vehicle's driving efficiency and safety.

[0004] In a first aspect, an embodiment of the present invention provides a method for identifying an obstacle, the method comprising: obtaining attribute information of a target obstacle on a current road; wherein the attribute information comprises at least one of a value attribute of the target obstacle, a degree of damage to the target obstacle when the target obstacle contacts a current vehicle, and a degree of damage to the current vehicle when the target obstacle contacts the current vehicle; the target obstacle contacts the current vehicle, including the current vehicle running over or colliding with the target obstacle; based on the attribute information, determining an estimation parameter of the target obstacle; based on the estimation parameter, determining an impact category of the target obstacle; wherein the impact category is used to indicate the degree of impact of the target obstacle on the driving route of the current vehicle.

[0005] Optionally, in a first implementation manner of the first aspect of the present invention, the step of obtaining the attribute information of the target obstacle on the current road includes: identifying the obstacle type of the target obstacle on the current road; based on a preset correspondence between the obstacle type and the attribute information, determining the attribute information corresponding to the obstacle type of the target obstacle; wherein the correspondence includes at least one obstacle type, at least one attribute information, and attribute values ​​of each attribute information corresponding to each obstacle type.

[0006] Optionally, in a second implementation method of the first aspect of the present invention, the step of obtaining the attribute information of the target obstacle on the current road includes: extracting feature information of the target obstacle on the current road; inputting the feature information of the target obstacle into a preset attribute estimation model, so as to determine the degree of similarity between the target obstacle and a specified type of obstacle based on the feature information of the target obstacle through the attribute estimation model, and outputting the attribute information of the target obstacle based on the degree of similarity.

[0007] Optionally, in a third implementation of the first aspect of the present invention, the above-mentioned attribute information includes the value attribute of the target obstacle, the degree of damage of the target obstacle when the target obstacle contacts the current vehicle, and the degree of damage of the current vehicle when the target obstacle contacts the current vehicle; the step of determining the valuation parameters of the target obstacle based on the attribute information includes: determining the damage value of the target obstacle based on the value attribute of the target obstacle and the degree of damage of the target obstacle; determining the damage value of the current vehicle based on the value parameters of the current vehicle and the degree of damage of the current vehicle; determining the valuation parameters of the target obstacle based on the damage value of the target obstacle and the damage value of the current vehicle.

[0008] Optionally, in a fourth implementation of the first aspect of the present invention, the above-mentioned attribute information also includes: the degree of attention of the target obstacle to the current vehicle; after the step of determining the valuation parameters of the target obstacle based on the damage value of the target obstacle and the damage value of the current vehicle, the method also includes: adjusting the valuation parameters of the target obstacle based on the degree of attention to obtain the final valuation parameters.

[0009] Optionally, in a fifth implementation of the first aspect of the present invention, the degree of attention of the target obstacle to the current vehicle is determined by: determining the motion attribute of the target obstacle, and if the motion attribute of the target obstacle is movable, obtaining the speed change information of the target obstacle; based on the speed change information, determining the degree of attention of the target obstacle to the current vehicle.

[0010] Optionally, in a sixth implementation manner of the first aspect of the present invention, the above-mentioned impact categories include: absolutely uncrushable, uncrushable, possibly uncrushable, possibly crushable and crushable; each impact category corresponds to a preset valuation parameter value range; based on the valuation parameter, the step of determining the impact category of the target obstacle includes: determining the valuation parameter value range to which the valuation parameter belongs, and determining the impact category corresponding to the valuation parameter value range to which the valuation parameter belongs as the impact category of the target obstacle.

[0011] Optionally, in the seventh implementation manner of the first aspect of the present invention, after the above-mentioned step of determining the impact category of the target obstacle based on the estimation parameters, the method further includes: determining the driving path of the current vehicle based on the impact category of the target obstacle; wherein the impact category of the obstacle passed by the driving path meets preset conditions.

[0012] In a second aspect, an embodiment of the present invention provides an obstacle identification device, which includes: an attribute information acquisition module, which acquires attribute information of a target obstacle on a current road; wherein the attribute information includes at least one of a value attribute of the target obstacle, a degree of damage to the target obstacle when the target obstacle contacts the current vehicle, and a degree of damage to the current vehicle when the target obstacle contacts the current vehicle; the contact between the target obstacle and the current vehicle includes the current vehicle running over or colliding with the target obstacle; an estimation parameter determination module, which is used to determine the estimation parameters of the target obstacle based on the attribute information; an impact category determination module, which is used to determine the impact category of the target obstacle based on the estimation parameters; wherein the impact category is used to indicate the degree of impact of the target obstacle on the driving route of the current vehicle.

[0013] In a third aspect, an embodiment of the present invention provides an electronic device, including a processor and a memory, wherein the memory stores machine executable instructions that can be executed by the processor, and the processor executes the machine executable instructions to implement the above-mentioned obstacle recognition method.

[0014] In a fourth aspect, an embodiment of the present invention provides a machine-readable storage medium, which stores machine-executable instructions. When the machine-executable instructions are called and executed by a processor, the machine-executable instructions prompt the processor to implement the above-mentioned obstacle recognition method.

[0015] The embodiments of the present invention bring the following beneficial effects:

[0016] The embodiment of the present invention provides an obstacle recognition method, device and electronic device, which first obtains the attribute information of the target obstacle on the current road; the attribute information includes at least one of the value attribute of the target obstacle, the damage degree of the target obstacle when the target obstacle contacts the current vehicle, and the damage degree of the current vehicle when the target obstacle contacts the current vehicle; then, based on the attribute information, the valuation parameter of the target obstacle is determined; finally, based on the valuation parameter, the impact category of the target obstacle is determined; wherein the impact category is used to indicate the impact degree of the target obstacle on the driving route of the current vehicle. This method estimates the obstacle through the attribute information of the obstacle, and then determines the impact degree of the obstacle on the driving route of the current vehicle, accurately determines the impact degree of the road obstacle on the driving route of the vehicle, and thus improves the driving efficiency and safety of the vehicle.

[0017] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention are realized and obtained by the structures particularly pointed out in the description, claims and drawings.

[0018] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without paying any creative work.

[0020] Figure 1 A flowchart of an obstacle recognition method provided by an embodiment of the present invention;

[0021] Figure 2 A flowchart of another obstacle recognition method provided by an embodiment of the present invention;

[0022] Figure 3 A schematic diagram of the structure of an obstacle recognition device provided by an embodiment of the present invention;

[0023] Figure 4 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0024] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0025] With the development of autonomous driving technology, the use of autonomous driving vehicles is becoming more and more widespread. In related technologies, when a vehicle detects obstacles on the road through autonomous driving technology, the default mode is to assume that all detected obstacles are non-collisionable and non-crushable, and plan the driving route to avoid all obstacles. However, this method may cause sudden braking or sharp turns on the driving route, affecting driving efficiency and riding experience.

[0026] Based on this, an embodiment of the present invention provides an obstacle recognition method, device and electronic device, and the technology can be applied to the obstacle recognition process in various traffic scenes.

[0027] To facilitate understanding of this embodiment, a method for identifying obstacles is first introduced. Figure 1 As shown, the method comprises the following steps:

[0028] Step 102, obtaining attribute information of the target obstacle on the current road; wherein the attribute information includes at least one of a value attribute of the target obstacle, a damage degree of the target obstacle when the target obstacle contacts the current vehicle, and a damage degree of the current vehicle when the target obstacle contacts the current vehicle; the contact of the target obstacle with the current vehicle includes the current vehicle running over or colliding with the target obstacle.

[0029] The target obstacles mentioned above may include multiple types of items, such as stones, flowers, garbage bags, small animals, pedestrians, other vehicles, etc. Different types of target obstacles have different attribute information. For example, the value attribute of a stone is low. When the stone contacts the current vehicle, the damage to the stone is low, but the damage to the current vehicle is high. The value attribute of a pedestrian is high. When the pedestrian contacts the current vehicle, the damage to the pedestrian is high, and the damage to the current vehicle may be medium. The value attribute of a target obstacle can be represented by a numerical value, and a higher numerical value can usually represent a higher value attribute; the damage degree of the target obstacle and the damage degree of the current vehicle can also be represented by a numerical value, and a higher numerical value can usually represent a higher degree of loss.

[0030] Since the attribute information of the target obstacle has a certain correspondence with its type, when acquiring the attribute information of the target obstacle, the basic type of the target obstacle, that is, what the target obstacle is, can be first identified through the sensor data acquired by the on-board sensor of the current vehicle. Among them, the above-mentioned on-board sensor can be an image sensor, and the collected sensor data is image data. Specifically, the image recognition technology can be used to identify the type of the obstacle; the above-mentioned on-board sensor can also be a laser sensor or an infrared sensor light. For the sensor data collected by these sensors, the corresponding data processing technology can be used to process the sensor data, so as to identify the type of the obstacle.

[0031] After the basic type of the target obstacle is identified, the attribute information of the target obstacle can be determined based on the predetermined correspondence between the type of the target obstacle and the attribute information. The correspondence between the type of obstacle and the attribute information can be determined based on historical experience or by relevant staff based on experiments. The correspondence can also be adjusted based on the collision effects between the vehicle and different types of obstacles during the driving process of the vehicle.

[0032] If the basic type of the target obstacle cannot be directly identified, the characteristic information of the target obstacle, such as the material, size, color, etc. of the target obstacle, can be extracted from the sensor data collected by the sensor, and the attribute information of the target obstacle can be determined based on the characteristic information. This process can be achieved through a pre-trained estimation model. The estimation model needs to be trained with a large amount of obstacle data.

[0033] Step 104: Determine the estimated parameters of the target obstacle based on the attribute information.

[0034] The above valuation parameters are mainly used to characterize the loss value caused if the current vehicle contacts the target obstacle, such as hitting the target obstacle or crushing the target obstacle. Since both the target obstacle and the current vehicle may suffer value loss when the current vehicle contacts the target obstacle, the valuation parameters can be calculated through two parts.

[0035] The loss value corresponding to the target obstacle can be determined based on the value attribute of the target obstacle and the degree of damage to the target obstacle after contact; the loss value corresponding to the current vehicle can be determined based on the value attribute of the current vehicle and the degree of damage to the current vehicle after contact, and finally the valuation parameter of the target obstacle is calculated by the loss value of the target obstacle and the current vehicle, such as taking the sum of the loss value of the target obstacle and the current vehicle as the valuation parameter of the target obstacle. Among them, the above-mentioned value attribute, degree of damage, loss value and valuation parameter can all be expressed by numerical values. When the corresponding numerical value of the valuation parameter is larger, the loss value caused by the current vehicle contacting the target obstacle is greater.

[0036] When the target obstacle is some creature with self-awareness, such as pedestrians, dogs, etc., they can avoid it when they see the vehicle coming towards them. It can be considered that the attribute information of these target obstacles also includes the degree of attention to the current vehicle. The degree of attention can be determined based on the change in the speed of the target obstacle when the current vehicle approaches the target obstacle. When the speed of the target obstacle becomes faster as the current vehicle gets closer to the target obstacle, it is considered that the target obstacle has a degree of attention to the current vehicle, and the faster the speed changes, the higher the degree of attention of the target obstacle. The valuation parameters can be adjusted based on the degree of attention of the target obstacle. For example, the attention coefficient of the valuation parameter can be calculated based on the degree of attention, and the valuation parameter can be multiplied by the attention coefficient to obtain the adjusted valuation parameter; the higher the degree of attention, the lower the adjusted valuation parameter.

[0037] Step 106 : determining the impact category of the target obstacle based on the estimated parameters; wherein the impact category is used to indicate the impact degree of the target obstacle on the driving route of the current vehicle.

[0038] The above-mentioned impact category can be predetermined, and multiple value ranges are determined according to the degree of impact of the target obstacle on the current vehicle's driving route, and each value range corresponds to an impact category. When the valuation parameter falls within a certain value range, the impact category of the target obstacle is the impact category corresponding to the value range. Since the valuation parameter can represent the loss value caused by the current vehicle contacting the target obstacle, the larger the valuation parameter of the target obstacle, the greater the degree of impact of the target obstacle indicated by its corresponding impact category on the current vehicle's driving route.

[0039] The embodiment of the present invention provides an obstacle recognition method, which first obtains the attribute information of the target obstacle on the current road; the attribute information includes at least one of the value attribute of the target obstacle, the damage degree of the target obstacle when the target obstacle contacts the current vehicle, and the damage degree of the current vehicle when the target obstacle contacts the current vehicle; then, based on the attribute information, the valuation parameter of the target obstacle is determined; finally, based on the valuation parameter, the impact category of the target obstacle is determined; wherein the impact category is used to indicate the impact degree of the target obstacle on the driving route of the current vehicle. This method estimates the obstacle through the attribute information of the obstacle, and then determines the impact degree of the obstacle on the driving route of the current vehicle, accurately determines the impact degree of the road obstacle on the driving route of the vehicle, and thus improves the driving efficiency and safety of the vehicle.

[0040] The following embodiment provides a specific implementation process for obtaining the attribute information of the target obstacle on the current road.

[0041] Since there is a certain correspondence between the attribute information of an obstacle and its type, when acquiring the attribute information of a target obstacle, the obstacle type of the target obstacle on the current road can be first identified; wherein the obstacle type can be understood as what the obstacle is; then based on the preset correspondence between the obstacle type and the attribute information, the attribute information corresponding to the obstacle type of the target obstacle is determined; wherein the correspondence includes at least one obstacle type, at least one attribute information, and the attribute value of each attribute information corresponding to each obstacle type. For example, when the category of the target obstacle is identified as a garbage bag, the preset correspondence is that the attribute value of the value attribute of the garbage bag is 0, the attribute value of the degree of damage of the garbage bag is 0, and the attribute value of the degree of damage of the current vehicle is 0, then the attribute values ​​corresponding to these attribute information can be assigned to the target obstacle.

[0042] When the obstacle type cannot be directly identified, the characteristic information of the target obstacle on the current road, such as size, shape, etc., can be extracted; the characteristic information of the target obstacle is input into a preset attribute estimation model to determine the degree of similarity between the target obstacle and the specified type of obstacle based on the characteristic information of the target obstacle through the attribute estimation model, and the attribute information of the target obstacle is output based on the similarity.

[0043] Among them, the above-mentioned specified types of obstacles may include multiple types, such as stones, garbage bags, pedestrians, etc. The attribute estimation model can compare the characteristic information of the target obstacle with the characteristics of the specified types of obstacles, so as to determine the similarity between the target obstacle and the specified types of obstacles. Further, the attribute information of the specified types of obstacles with a higher degree of similarity with the target obstacle can be selected as the attribute information of the target obstacle, or the attribute information of multiple specified types of obstacles with a degree of similarity with the target obstacle higher than a preset threshold can be weightedly calculated, and the calculation result is used as the attribute information of the target obstacle.

[0044] The embodiment of the present invention also provides another obstacle recognition method. Figure 1 This method mainly introduces the specific implementation process of determining the estimation parameters of the target obstacle and the specific implementation process of determining the impact category of the target obstacle. Figure 2 As shown, the method comprises the following steps:

[0045] Step 202: Obtain attribute information of target obstacles on the current road.

[0046] The above-mentioned attribute information may include the value attribute of the target obstacle, the damage degree of the target obstacle when the target obstacle contacts the current vehicle, and the damage degree of the current vehicle when the target obstacle contacts the current vehicle.

[0047] Step 204: Determine the damage value of the target obstacle based on the value attribute of the target obstacle and the damage degree of the target obstacle. Specifically, the attribute value corresponding to the value attribute of the target obstacle and the attribute value corresponding to the damage degree of the target obstacle may be multiplied, and the multiplication result is used as the damage value of the target obstacle.

[0048] Step 206, based on the value parameter of the current vehicle and the damage degree of the current vehicle, determine the damage value of the current vehicle. Specifically, the parameter value corresponding to the value parameter of the current vehicle can be multiplied by the attribute value corresponding to the damage degree of the current vehicle, and the multiplication result is used as the damage value of the current vehicle. In addition, other forms of calculation can be performed based on the value attribute (or value parameter) and the damage degree to obtain the damage value. The specific form is not limited, but it needs to be ensured that: for the same damage degree, the higher the value attribute, the greater the corresponding damage value; for the same value attribute, the higher the damage degree, the greater the corresponding damage value.

[0049] Step 208: Determine the valuation parameter of the target obstacle based on the damage value of the target obstacle and the damage value of the current vehicle. Specifically, the damage value of the target obstacle and the damage value of the current vehicle may be added, and the added result is used as the valuation parameter of the target obstacle.

[0050] Step 210: Based on the degree of attention, the estimation parameters of the target obstacle are adjusted to obtain the final estimation parameters.

[0051] When the target obstacle is a conscious creature, such as a pedestrian, the target obstacle can notice the current vehicle and avoid it. At this time, the attribute information of the target obstacle can also include the degree of attention of the target obstacle to the current vehicle.

[0052] The degree of attention of the above target obstacles to the current vehicle can be determined in the following ways:

[0053] (1) Determine the motion attribute of the target obstacle. If the motion attribute of the target obstacle is movable, obtain the speed change information of the target obstacle.

[0054] Specifically, the position of the target obstacle at different times can be determined through the sensor data collected multiple times by the on-board sensors, thereby determining whether the target obstacle can move; if it can move, its speed change is calculated based on the position of the target obstacle at different times.

[0055] (2) Based on the speed change information, determine the degree of attention of the target obstacle to the current vehicle.

[0056] If the speed change information of the target obstacle indicates that the speed of the target obstacle increases as the current vehicle approaches, it can be considered that the target obstacle pays attention to the current vehicle, and the higher the speed increases, the higher the degree of attention. If the speed change information of the target obstacle indicates that the speed of the target obstacle remains unchanged or slows down as the current vehicle approaches, it can be considered that the target obstacle does not pay attention to the current vehicle, and the degree of attention is zero.

[0057] When the attention level is not zero, the parameter value corresponding to the attention level and the estimated parameter of the target obstacle can be used for calculation, so that the estimated parameter of the target obstacle decreases as the attention level increases. For example, the value range of the attention level can be [0,1), and when the estimated parameter is adjusted using the attention level, the following calculation formula can be used:

[0058] Adjusted valuation parameter = (1-degree of concern) × initial valuation parameter.

[0059] Step 212, determining the valuation parameter value range to which the valuation parameter belongs, and determining the impact category corresponding to the valuation parameter value range to which the valuation parameter belongs as the impact category of the target obstacle.

[0060] In specific implementation, the impact categories can be set as follows: absolutely not crushable, not crushable, possibly not crushable, possibly crushable, and crushable; each impact category corresponds to a preset valuation parameter value range. For example, the impact category of pedestrians can be absolutely not crushable, and the impact category of flowers and plants can be possibly crushable. The valuation parameters of different target obstacles fall within the valuation parameter value range corresponding to their corresponding impact categories.

[0061] Step 214, based on the impact category of the target obstacle, determine the driving path of the current vehicle; wherein the impact category of the obstacle passed by the driving path meets a preset condition.

[0062] The above preset conditions can be obstacles of certain impact categories, such as the impact categories of possible crushable and crushable, that is, the current vehicle's driving path can pass through obstacles that are possible crushable and crushable. This method makes the vehicle more flexible when planning the driving path, better balances the relationship between driving efficiency and contact with obstacles, and takes into account the riding experience of the passengers on the vehicle.

[0063] The embodiment of the present invention also provides another obstacle recognition method. Figure 1 This method is mainly aimed at the following road scenes to determine the crushability attribute classification of obstacles, so that the unmanned vehicle can distinguish the crushability level of obstacles, and thus make the most reasonable (least cost) driving road planning.

[0064] Scenario 1: When garbage debris on the road is detected as an obstacle (such as plastic bags and flying paper), experienced human drivers will choose to drive slowly over it, while bricks, scattered furniture, and even vehicles, pedestrians, and road facilities on the road need to be detoured. At this time, it is necessary to distinguish which obstacles can be crushed and which cannot be crushed.

[0065] Scenario 2: When obstacles on the road are not crushable, but the vehicle speed does not allow the car to stop completely, the autonomous vehicle (also called "driverless car") needs to classify the crushability / collisionability of obstacles (people, cars, vegetation, small animals, traffic facilities, etc.) to minimize the cost of hitting the obstacles, such as people and cars, people and vegetation, people and small animals, traffic cones and small animals, etc.

[0066] Scenario 3: When there are many dynamic obstacles on the road and it is very crowded, such as many pedestrians and vehicles in a village in the city or sheep on a rural road, most dynamic obstacles cannot be crushed. However, they may be able to sense and avoid the approach of the unmanned vehicle. If the unmanned vehicle needs to pass the road slowly, it needs to judge the perception and avoidance capabilities of these obstacles and dynamically modify the crushability of these obstacles, so that a route through the crowded road can be planned.

[0067] By refining the properties of obstacles and the degree of damage caused to the unmanned vehicle, an empirical formula for the crushability of obstacles can be established, thereby calculating the obstacle crushability index.

[0068] This method needs to be implemented through the following steps:

[0069] 1. Classification of obstacle attributes.

[0070] These attributes can include the following:

[0071] 1. The value of the obstacle itself (equivalent to the above-mentioned "value attribute of the target obstacle"), represented by V (Value): whether the obstacle itself has high value. It can be assumed that the value range of its own value is 0-3. For example, the value of a pedestrian is 3, while the value of a garbage bag is 0.

[0072] 2. Damage from rolling over (equivalent to the above-mentioned "damage level of the target obstacle"), expressed as D (Damage): How much damage will be caused to the obstacle if it is rolled over / hit. The damage level can be classified from 0-3, from no damage to complete damage. For example, a separation wall may be completely intact (0), while hitting a pedestrian may cause the pedestrian's death (3).

[0073] 3. The damage of the unmanned vehicle (equivalent to the above-mentioned "current vehicle damage level"), expressed as De (Ego-cardamage): how much damage will the unmanned vehicle cause when it runs over / collides with an obstacle. The damage level can be classified into 0-3, from no damage to complete damage. For example, a separation wall may cause the unmanned vehicle to be completely destroyed (3), while running over a plastic bag will have no effect on the unmanned vehicle (0).

[0074] 4. Attention level, represented by A (Attentiveness): whether the obstacle is paying attention to the state of the unmanned vehicle. Most obstacles are static, while people, cars, animals, etc. may pay attention to the state of the unmanned vehicle and make game judgments. Specifically, it is possible to judge whether the obstacle will respond to the unmanned vehicle by analyzing the image and movement dynamics of the obstacle. If the obstacle's attention level is 1, the obstacle's crushable attribute will be dynamically adjusted, so that the unmanned vehicle can plan a path, approach the obstacle, and make real-time judgments on whether the obstacle is paying attention and making evasive responses, so that the unmanned vehicle can autonomously interact with dynamic obstacles on the road when passing through crowded roads.

[0075] (ii) Calculation of the crushability of obstacles (equivalent to the above-mentioned “impact category”).

[0076] Generally speaking, the crushability of obstacles can be divided into the following categories:

[0077] 1. Crustable (C): The obstacle has low value and will not cause damage to the autonomous vehicle. This category may include soft small obstacles, i.e. empty plastic bags, low vegetation (grass), paper scraps, etc.

[0078] 2. Possibly crushable (LC): obstacles have a lower value, but may have some impact on the autonomous vehicle. These categories may include small solid obstacles, i.e. shallow potholes, bottles of various materials. In this case, it may be OK to crush these obstacles at low speeds, and it will only make passengers feel a little bumpy.

[0079] 3. Likely Not Crushable (LNC): The detected obstacle has a certain value and may cause some damage to the autonomous vehicle. The category may include road obstacles.

[0080] 4. Non-Crushable (NC): The obstacle has a high value and collision will cause damage to the obstacle and the autonomous vehicle. The category may include vehicles.

[0081] 5. Never Run Over (HNC): The detected obstacle has a high value and running over it may cause complete damage. The category may include pedestrians.

[0082] Assuming that the value of the driverless car is Ve (Ego-car value), based on the above analysis, we can classify the attributes of the driverless car and derive the following crashing cost CC (Crashing Cost):

[0083] CC=(V*D+Ve*De)*(1-A)

[0084] In determining the crushability of an obstacle, the higher the crushability estimate, the lower the crushability of the obstacle.

[0085] In addition, the above process of determining the crushability of obstacles may be directly regarded as a comprehensive classification task, thereby skipping the above empirical formula and learning a deep classification model through a large amount of data calibration and data-driven methods.

[0086] If the above formula is used to calculate the crushability of an obstacle, assuming that the value of the unmanned vehicle Ev=2, and the obstacle's attention to the unmanned vehicle is zero (A=0), the following Table 1 can be used to verify the empirical formula.

[0087] Table 1

[0088]

[0089] In principle, in scenario 1, crushability is a binary classification problem, that is, all obstacles that are not in the "crushable" category (CC>0) are not crushable and need to be avoided. In scenario 2, the unmanned vehicle needs to classify the crushability of all obstacles based on the crushing estimate, and make a crushing / collision route with the lowest crushing estimate in a short time. In scenario 3, by analyzing the degree of attention and response of the obstacle to the unmanned vehicle (A=0 or A=1), the unmanned vehicle can modify its crushability according to the degree of attention of the obstacle, thereby choosing a reasonable road to approach the obstacle, and according to the response of the obstacle, update the obstacle's attention estimate and crushability at any time, so as to dynamically adjust the approach route.

[0090] The above method proposes an empirical formula for calculating the cost estimation of the crushability of an obstacle, and based on this formula, evaluates the value of the obstacle in different scenarios, the damage caused by crushing / collision to the obstacle and the unmanned vehicle, and the perceived attention to the obstacle, thereby comprehensively estimating the crushability category of the obstacle.

[0091] Corresponding to the above method embodiment, the present invention provides an obstacle recognition device, such as Figure 3 As shown, the device comprises:

[0092] The attribute information acquisition module 302 acquires attribute information of a target obstacle on the current road; wherein the attribute information includes at least one of a value attribute of the target obstacle, a damage degree of the target obstacle when the target obstacle contacts the current vehicle, and a damage degree of the current vehicle when the target obstacle contacts the current vehicle; the contact between the target obstacle and the current vehicle includes the current vehicle running over or colliding with the target obstacle;

[0093] An estimation parameter determination module 304 is used to determine an estimation parameter of a target obstacle based on the attribute information;

[0094] The impact category determination module 306 is used to determine the impact category of the target obstacle based on the estimated parameters; wherein the impact category is used to indicate the impact degree of the target obstacle on the current vehicle's driving route.

[0095] Furthermore, the attribute information acquisition module is also used to: identify the obstacle type of the target obstacle on the current road; determine the attribute information corresponding to the obstacle type of the target obstacle based on a preset correspondence between the obstacle type and the attribute information; wherein the correspondence includes at least one obstacle type, at least one attribute information, and an attribute value of each attribute information corresponding to each obstacle type.

[0096] Furthermore, the above-mentioned attribute information acquisition module is also used to: extract characteristic information of the target obstacle on the current road; input the characteristic information of the target obstacle into a preset attribute estimation model, so as to determine the similarity between the target obstacle and the specified type of obstacle based on the characteristic information of the target obstacle through the attribute estimation model, and output the attribute information of the target obstacle based on the similarity.

[0097] Furthermore, the above-mentioned attribute information includes the value attribute of the target obstacle, the damage degree of the target obstacle when the target obstacle contacts the current vehicle, and the damage degree of the current vehicle when the target obstacle contacts the current vehicle; the above-mentioned valuation parameter determination module is also used to: determine the damage value of the target obstacle based on the value attribute of the target obstacle and the damage degree of the target obstacle; determine the damage value of the current vehicle based on the value parameters of the current vehicle and the damage degree of the current vehicle; determine the valuation parameters of the target obstacle based on the damage value of the target obstacle and the damage value of the current vehicle.

[0098] Furthermore, the attribute information also includes: the degree of attention of the target obstacle to the current vehicle; the device also includes: an estimation parameter adjustment module, which is used to adjust the estimation parameters of the target obstacle based on the degree of attention to obtain the final estimation parameters.

[0099] Furthermore, the above-mentioned device also includes an attention degree determination module, which is used to: determine the motion attribute of the target obstacle, if the motion attribute of the target obstacle is movable, obtain the speed change information of the target obstacle; based on the speed change information, determine the attention degree of the target obstacle to the current vehicle.

[0100] Furthermore, the above-mentioned impact categories include: absolutely uncrushable, uncrushable, possibly uncrushable, possibly crushable and crushable; each impact category corresponds to a preset valuation parameter value range; the above-mentioned impact category determination module is also used to: determine the valuation parameter value range to which the valuation parameter belongs, and determine the impact category corresponding to the valuation parameter value range to which the valuation parameter belongs as the impact category of the target obstacle.

[0101] Furthermore, the above-mentioned device also includes a driving path determination module, which is used to: determine the driving path of the current vehicle based on the impact category of the target obstacle; wherein the impact category of the obstacle passed by the driving path meets a preset condition.

[0102] This embodiment also provides an electronic device, including a processor and a memory, wherein the memory stores machine executable instructions that can be executed by the processor, and the processor executes the machine executable instructions to implement the above obstacle recognition method.

[0103] See also Figure 4 As shown, the electronic device includes a processor 100 and a memory 101. The memory 101 stores machine executable instructions that can be executed by the processor 100. The processor 100 executes the machine executable instructions to implement the above obstacle recognition method.

[0104] Further, Figure 4 The electronic device shown further includes a bus 102 and a communication interface 103 , and the processor 100 , the communication interface 103 and the memory 101 are connected via the bus 102 .

[0105] The memory 101 may include a high-speed random access memory (RAM), and may also include a non-volatile memory, such as at least one disk storage. The communication connection between the system network element and at least one other network element is realized through at least one communication interface 103 (which may be wired or wireless), and the Internet, wide area network, local area network, metropolitan area network, etc. may be used. The bus 102 may be an ISA bus, a PCI bus, or an EISA bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 4 Only one bidirectional arrow is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0106] The processor 100 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the hardware integrated logic circuit or software instructions in the processor 100. The above processor 100 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The methods, steps and logic block diagrams disclosed in the embodiments of the present invention can be implemented or executed. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor, etc. The steps of the method disclosed in conjunction with the embodiments of the present invention can be directly embodied as a hardware decoding processor for execution, or a combination of hardware and software modules in the decoding processor for execution. The software module may be located in a storage medium mature in the art, such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory 101, and the processor 100 reads the information in the memory 101 and completes the steps of the method of the above embodiment in combination with its hardware.

[0107] This embodiment further provides a machine-readable storage medium, which stores machine-executable instructions. When the machine-executable instructions are called and executed by a processor, the machine-executable instructions prompt the processor to implement the above obstacle recognition method.

[0108] An obstacle recognition method, device and electronic device provided in an embodiment of the present invention include a computer-readable storage medium storing program code, and the instructions included in the program code can be used to execute the method described in the previous method embodiment. The specific implementation can be referred to the method embodiment, which will not be repeated here.

[0109] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the system and device described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0110] In addition, in the description of the embodiments of the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, and it can be the internal communication of two components. For those skilled in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0111] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0112] In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc., indicating the orientation or positional relationship, are based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first", "second", and "third" are used for descriptive purposes only, and cannot be understood as indicating or implying relative importance.

[0113] Finally, it should be noted that the above embodiments are only specific implementations of the present invention, which are used to illustrate the technical solutions of the present invention, rather than to limit them. The protection scope of the present invention is not limited thereto. Although the present invention is described in detail with reference to the above embodiments, those skilled in the art should understand that any person skilled in the art can still modify the technical solutions recorded in the above embodiments within the technical scope disclosed by the present invention, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A method for identifying an obstacle, characterized in that: The method comprises: Acquiring attribute information of a target obstacle on the current road; wherein the attribute information includes a value attribute of the target obstacle, a damage degree of the target obstacle when the target obstacle contacts the current vehicle, and a damage degree of the current vehicle when the target obstacle contacts the current vehicle; the contact between the target obstacle and the current vehicle includes the current vehicle running over or colliding with the target obstacle; Determining, based on the attribute information, an estimation parameter of the target obstacle; Based on the estimated parameters, determining the impact category of the target obstacle; wherein the impact category is used to indicate the degree of impact of the target obstacle on the driving route of the current vehicle; The attribute information also includes: the degree of attention of the target obstacle to the current vehicle; The step of determining the estimated parameters of the target obstacle based on the attribute information includes: Determining the damage value of the target obstacle based on the value attribute of the target obstacle and the damage degree of the target obstacle; Determining the damage value of the current vehicle based on the value parameter of the current vehicle and the damage degree of the current vehicle; Determining an estimation parameter of the target obstacle based on the damage value of the target obstacle and the damage value of the current vehicle; Based on the degree of attention, adjusting the estimation parameters of the target obstacle to obtain final estimation parameters; The degree of attention of the target obstacle to the current vehicle is determined by: Determine the motion attribute of the target obstacle, and if the motion attribute of the target obstacle is movable, obtain speed change information of the target obstacle; Based on the speed change information, the degree of attention of the target obstacle to the current vehicle is determined.

2. The method according to claim 1, characterized in that The steps of obtaining the attribute information of the target obstacle on the current road include: Identify the obstacle type of the target obstacle on the current road; Determining the attribute information corresponding to the obstacle type of the target obstacle based on a preset correspondence between the obstacle type and the attribute information; The corresponding relationship includes at least one obstacle type, at least one attribute information, and an attribute value of each attribute information corresponding to each obstacle type.

3. The method according to claim 1, characterized in that The steps of obtaining the attribute information of the target obstacle on the current road include: Extract characteristic information of target obstacles on the current road; The characteristic information of the target obstacle is input into a preset attribute estimation model, so that the attribute estimation model can determine the similarity between the target obstacle and a specified type of obstacle based on the characteristic information of the target obstacle, and output the attribute information of the target obstacle based on the similarity.

4. The method according to claim 1, characterized in that: The impact categories include: absolutely not crushable, not crushable, possibly not crushable, possibly crushable, and crushable; each of the impact categories corresponds to a preset range of valuation parameter values; The step of determining the impact category of the target obstacle based on the estimated parameters comprises: The estimation parameter value range to which the estimation parameter belongs is determined, and the impact category corresponding to the estimation parameter value range to which the estimation parameter belongs is determined as the impact category of the target obstacle.

5. The method according to claim 1, characterized in that After the step of determining the impact category of the target obstacle based on the estimated parameters, the method further includes: Based on the impact category of the target obstacle, the driving path of the current vehicle is determined; wherein the impact category of the obstacle passed by the driving path meets a preset condition.

6. An obstacle recognition device, characterized in that: The device comprises: an attribute information acquisition module, which acquires attribute information of a target obstacle on the current road; wherein the attribute information includes a value attribute of the target obstacle, a damage degree of the target obstacle when the target obstacle contacts the current vehicle, and a damage degree of the current vehicle when the target obstacle contacts the current vehicle; the contact between the target obstacle and the current vehicle includes the current vehicle running over or colliding with the target obstacle; An estimation parameter determination module, used to determine the estimation parameters of the target obstacle based on the attribute information; An impact category determination module, used to determine the impact category of the target obstacle based on the estimated parameters; wherein the impact category is used to indicate the degree of impact of the target obstacle on the driving route of the current vehicle; The attribute information also includes: the degree of attention of the target obstacle to the current vehicle; The estimated parameter determination module is also used for: Determining the damage value of the target obstacle based on the value attribute of the target obstacle and the damage degree of the target obstacle; Determining the damage value of the current vehicle based on the value parameter of the current vehicle and the damage degree of the current vehicle; Determining an estimation parameter of the target obstacle based on the damage value of the target obstacle and the damage value of the current vehicle; Based on the degree of attention, adjusting the estimation parameters of the target obstacle to obtain final estimation parameters; The degree of attention of the target obstacle to the current vehicle is determined by: Determine the motion attribute of the target obstacle, and if the motion attribute of the target obstacle is movable, obtain speed change information of the target obstacle; Based on the speed change information, the degree of attention of the target obstacle to the current vehicle is determined.

7. An electronic device, characterized in that: The invention comprises a processor and a memory, wherein the memory stores machine executable instructions that can be executed by the processor, and the processor executes the machine executable instructions to implement the obstacle recognition method according to any one of claims 1 to 5.

8. A machine-readable storage medium, characterized in that: The machine-readable storage medium stores machine-executable instructions. When the machine-executable instructions are called and executed by a processor, the machine-executable instructions prompt the processor to implement the obstacle recognition method according to any one of claims 1 to 5.

Images