CN110349425B - An important target generation method for vehicle-road cooperative autonomous driving system - Google Patents
An important target generation method for vehicle-road cooperative autonomous driving systemDownload PDFInfo
- Publication number
- CN110349425B CN110349425BCN201910745048.8ACN201910745048ACN110349425BCN 110349425 BCN110349425 BCN 110349425BCN 201910745048 ACN201910745048 ACN 201910745048ACN 110349425 BCN110349425 BCN 110349425B
- Authority
- CN
- China
- Prior art keywords
- information
- target
- degree
- behavior
- calculating
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/09—Arrangements for giving variable traffic instructions
- G08G1/0962—Arrangements for giving variable traffic instructions having an indicator mounted inside the vehicle, e.g. giving voice messages
- G08G1/0967—Systems involving transmission of highway information, e.g. weather, speed limits
- G08G1/096708—Systems involving transmission of highway information, e.g. weather, speed limits where the received information might be used to generate an automatic action on the vehicle control
- G08G1/096725—Systems involving transmission of highway information, e.g. weather, speed limits where the received information might be used to generate an automatic action on the vehicle control where the received information generates an automatic action on the vehicle control
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/16—Anti-collision systems
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Atmospheric Sciences (AREA)
- Traffic Control Systems (AREA)
Abstract
Description
Translated fromChinese技术领域technical field
本发明涉及自动驾驶领域，特别是涉及一种用于车路协同自动驾驶系统的重要目标生成方法。The invention relates to the field of automatic driving, in particular to an important target generation method for a vehicle-road cooperative automatic driving system.
背景技术Background technique
自动驾驶技术是当前汽车产业的技术热点，根据SAE的自动驾驶分级，目前主要划分为L0-L5这六个自动驾驶分级，其中L0级指没有任何自动驾驶功能的车辆，L1-L2级自动驾驶本质上仍是驾驶辅助系统(ADAS)，L3级自动驾驶可以称之为准自动驾驶系统，L4-L5级自动驾驶可以认为是真正有意义的自动驾驶系统。Autopilot technology is a technical hotspot in the current automobile industry. According to the autopilot classification of SAE, it is currently mainly divided into six autopilot classifications, L0-L5, of which L0 refers to vehicles without any autopilot function, L1-L2 autopilot In essence, it is still a driver assistance system (ADAS). L3 autonomous driving can be called a quasi-autonomous driving system, and L4-L5 autonomous driving can be considered as a truly meaningful autonomous driving system.
传统的L1-L2级别自动驾驶车辆，主要利用车载传感器(GPS、IMU、轮速传感器等)和感知传感器(前向雷达、前视摄像头、超声波雷达等)实现简单场景的辅助驾驶功能，例如ACC(Adaptive Cruise Control)，AEB，TJA，HWA等。随着自动驾驶功能和安全等级的提升，车辆需要具有更精确的感知定位能力，更加可靠稳定的决策控制能力，能够处理更复杂的场景。因此对于自车和周边环境感知能力提出了更高的要求，例如L3-L5的自动驾驶车辆通过增加前向激光雷达、多个角雷达和侧雷达、高像素前视摄像头、侧视摄像头、后视摄像头、高精度地图服务器等，来实现高精度地图/定位、动静态目标检测跟踪、车道路沿检测、交通标识识别等环境感知能力。Traditional L1-L2 level autonomous vehicles mainly use on-board sensors (GPS, IMU, wheel speed sensors, etc.) and perception sensors (forward radar, forward-looking camera, ultrasonic radar, etc.) to achieve assisted driving functions in simple scenarios, such as ACC (Adaptive Cruise Control), AEB, TJA, HWA, etc. With the improvement of automatic driving functions and safety levels, vehicles need to have more accurate perception and positioning capabilities, more reliable and stable decision-making control capabilities, and can handle more complex scenarios. Therefore, higher requirements are put forward for the perception ability of the self-vehicle and the surrounding environment. For example, L3-L5 autonomous vehicles can increase the forward-facing lidar, multiple angle radars and side radars, high-pixel front-view cameras, side-view cameras, and rear-view cameras. Video cameras, high-precision map servers, etc., to achieve high-precision map/positioning, dynamic and static target detection and tracking, road edge detection, traffic sign recognition and other environmental perception capabilities.
车路协同系统中，车辆或路基设施可以通过感知传感器获取周边目标信息，其中大部分目标是普通目标，简单而言就是不会对本车整车行驶产生很大影响的目标；有些目标是重要目标，简单而言就是会有很大可能影响本车正常行驶的目标。对于重要目标，自动驾驶车辆需要重点监测以获取其位置、速度、加速度、航向角等信息，使传感器感知能力集中的方向同步跟踪锁定重要目标。因此需要对该区域内的目标进行较为细致的分析，以获取重要目标的列表。In the vehicle-road coordination system, vehicles or roadbed facilities can obtain surrounding target information through sensing sensors. Most of the targets are ordinary targets, in short, targets that will not have a great impact on the driving of the vehicle; some targets are important targets. , in short, it is likely to affect the normal driving target of the vehicle. For important targets, autonomous vehicles need to focus on monitoring to obtain information such as their position, speed, acceleration, heading angle, etc., so that the direction with concentrated sensor perception capabilities can synchronously track and lock important targets. Therefore, a more detailed analysis of the targets in this area is required to obtain a list of important targets.
发明内容SUMMARY OF THE INVENTION
本发明的一个目的是提供一种用于车路协同自动驾驶系统的重要目标生成方法，能够对车辆周边的目标进行细致的分析以获取重要目标列表。An object of the present invention is to provide an important target generation method for a vehicle-road cooperative automatic driving system, which can perform detailed analysis on the targets around the vehicle to obtain a list of important targets.
特别地，本发明提供了一种用于车路协同自动驾驶系统的重要目标生成方法，包括以下步骤：In particular, the present invention provides an important target generation method for a vehicle-road cooperative automatic driving system, comprising the following steps:
获取所述车辆周边的多个目标的相关信息；acquiring relevant information of multiple targets around the vehicle;
根据所述相关信息依次计算每一所述目标的行为特征信息和属性特征信息；Calculate the behavior feature information and attribute feature information of each of the targets in turn according to the relevant information;
根据所述行为特征信息和所述属性特征信息计算所述目标的整体重要度；Calculate the overall importance of the target according to the behavior feature information and the attribute feature information;
将所述整体重要度大于第一阈值的所述目标按照所述整体重要度的大小进行降序排列，以形成待选列表；arranging the targets whose overall importance is greater than the first threshold in descending order according to the size of the overall importance to form a candidate list;
在所述待选列表中从最前项依次往后选取不超过预设数量的所述目标，以形成重要目标列表。In the to-be-selected list, the targets not exceeding a preset number are selected sequentially from the top item to the back to form an important target list.
可选地，所述行为特征信息包括：所述目标的行为异常度、行为鲁莽度和碰撞风险度；Optionally, the behavior feature information includes: behavior abnormality degree, behavior recklessness degree and collision risk degree of the target;
所述属性特征信息包括：所述目标的自身脆弱度、重大威胁度和额外兴趣度。The attribute feature information includes: the target's own vulnerability, major threat and extra interest.
可选地，根据所述行为特征信息和所述属性特征信息计算所述目标的整体重要度，包括：Optionally, calculating the overall importance of the target according to the behavior feature information and the attribute feature information, including:
对所述目标的所述行为异常度、所述行为鲁莽度、所述碰撞风险度、所述自身脆弱度、所述重大威胁度和所述额外兴趣度进行加权并取其中的最大值作为所述目标的所述整体重要度。The abnormality degree of behavior, the degree of recklessness of behavior, the degree of collision risk, the degree of self-vulnerability, the degree of major threat and the degree of extra interest of the target are weighted, and the maximum value is taken as the the overall importance of the target.
可选地，对所述目标的所述行为异常度、所述行为鲁莽度、所述碰撞风险度、所述自身脆弱度、所述重大威胁度和所述额外兴趣度进行加权并取其中的最大值作为所述目标的所述整体重要度，包括：Optionally, the abnormality degree of behavior, the degree of recklessness of behavior, the degree of collision risk, the degree of self-vulnerability, the degree of major threat, and the degree of extra interest of the target are weighted and take the The maximum value is used as the overall importance of the target, including:
所述碰撞风险度和所述额外兴趣度的加权系数均大于所述行为异常度、所述行为鲁莽度、所述自身脆弱度或所述重大威胁度的加权系数。The weighting coefficients of the collision risk degree and the additional interest degree are both larger than the weighting coefficients of the behavior abnormality degree, the behavior reckless degree, the self-vulnerability degree or the major threat degree.
可选地，根据所述相关信息依次计算每一所述目标的行为特征信息和属性特征信息，包括：Optionally, sequentially calculating the behavior feature information and attribute feature information of each of the targets according to the relevant information, including:
获取所述目标当前位置的交通规则信息；Obtain the traffic rule information of the current location of the target;
获取所述目标当前的交通行为信息；obtain the current traffic behavior information of the target;
根据所述交通规则信息和所述交通行为信息计算所述行为异常度。The behavior abnormality degree is calculated according to the traffic rule information and the traffic behavior information.
可选地，根据所述相关信息依次计算每一所述目标的行为特征信息和属性特征信息，包括：Optionally, sequentially calculating the behavior feature information and attribute feature information of each of the targets according to the relevant information, including:
获取所述目标的运动信息变化率信息；obtaining the rate of change information of the motion information of the target;
根据所述运动信息变化率信息计算所述行为鲁莽度。The behavior recklessness is calculated according to the motion information change rate information.
可选地，根据所述相关信息依次计算每一所述目标的行为特征信息和属性特征信息，包括：Optionally, sequentially calculating the behavior feature information and attribute feature information of each of the targets according to the relevant information, including:
计算所述目标与所述车辆发生碰撞的风险值；calculating a risk value of the collision between the target and the vehicle;
根据所述风险值计算所述碰撞风险度。The collision risk is calculated according to the risk value.
可选地，根据所述相关信息依次计算每一所述目标的行为特征信息和属性特征信息，包括：Optionally, sequentially calculating the behavior feature information and attribute feature information of each of the targets according to the relevant information, including:
获取所述目标的分类信息、外观信息和尺寸信息；Obtain classification information, appearance information and size information of the target;
根据所述分类信息、所述外观信息和所述尺寸信息计算所述自身脆弱度。The self vulnerability is calculated based on the classification information, the appearance information and the size information.
可选地，根据所述相关信息依次计算每一所述目标的行为特征信息和属性特征信息，包括：Optionally, sequentially calculating the behavior feature information and attribute feature information of each of the targets according to the relevant information, including:
获取所述目标的分类信息、外观信息、尺寸信息和外挂信息；Obtain the classification information, appearance information, size information and plug-in information of the target;
根据所述分类信息、所述外观信息、所述尺寸信息和所述外挂信息计算所述重大威胁度。The major threat degree is calculated according to the classification information, the appearance information, the size information and the plug-in information.
可选地，根据所述相关信息依次计算每一所述目标的行为特征信息和属性特征信息，包括：Optionally, sequentially calculating the behavior feature information and attribute feature information of each of the targets according to the relevant information, including:
接收第三方利益相关者发送的需要额外关注的目标列表；Receive a list of targets that require additional attention from third-party stakeholders;
根据所述目标列表计算所述额外兴趣度。The additional interest degree is calculated according to the target list.
本发明根据目标的行为特征信息和属性特征信息分析其是否对本车具有重大影响，即计算整体重要度，并根据影响的大小(即整体重要度的大小)选出一定数量的重要目标，并形成降序排列的重要目标列表，以供车辆后续分析计算使用。本方法通过对车辆周边的目标进行细致的分析，获取到该重要目标列表。The present invention analyzes whether the target has a significant impact on the vehicle according to the behavior feature information and attribute feature information of the target, that is, calculates the overall importance, and selects a certain number of important targets according to the size of the impact (that is, the overall importance), and forms A list of important targets in descending order for use in subsequent vehicle analysis calculations. The method obtains the list of important targets by analyzing the targets around the vehicle in detail.
进一步地，本发明提出了一种新的方法获取目标的重要度细分度量衡，包括对目标的行为异常度、行为鲁莽度、碰撞风险度、目标的自身脆弱度、重大威胁度和额外兴趣度六个度量衡的计算，最后通过加权平均的方式获得目标的重要度，最后输出重要度高的目标形成重要目标列表。Further, the present invention proposes a new method to obtain the subdivision measures of the importance of the target, including the abnormality of the target's behavior, the degree of reckless behavior, the risk of collision, the target's own vulnerability, the degree of major threat, and the degree of extra interest. The six metrics are calculated, and finally the importance of the target is obtained by weighted average, and finally the target with high importance is output to form an important target list.
根据下文结合附图对本发明具体实施例的详细描述，本领域技术人员将会更加明了本发明的上述以及其他目的、优点和特征。The above and other objects, advantages and features of the present invention will be more apparent to those skilled in the art from the following detailed description of the specific embodiments of the present invention in conjunction with the accompanying drawings.
附图说明Description of drawings
后文将参照附图以示例性而非限制性的方式详细描述本发明的一些具体实施例。附图中相同的附图标记标示了相同或类似的部件或部分。本领域技术人员应该理解，这些附图未必是按比例绘制的。附图中：Hereinafter, some specific embodiments of the present invention will be described in detail by way of example and not limitation with reference to the accompanying drawings. The same reference numbers in the figures designate the same or similar parts or parts. It will be understood by those skilled in the art that the drawings are not necessarily to scale. In the attached picture:
图1是根据本发明一个实施例的重要目标生成方法的流程框图。FIG. 1 is a flowchart of a method for generating an important target according to an embodiment of the present invention.
具体实施方式Detailed ways
图1是根据本发明一个实施例的重要目标生成方法的流程框图。如图1所示，本发明提供了一种用于车路协同自动驾驶系统的重要目标生成方法，其一般性地可以包括以下步骤：FIG. 1 is a flowchart of a method for generating an important target according to an embodiment of the present invention. As shown in FIG. 1 , the present invention provides an important target generation method for a vehicle-road cooperative automatic driving system, which generally includes the following steps:
S10：获取车辆周边的多个目标的相关信息。S10: Acquire relevant information of multiple targets around the vehicle.
该相关信息可以包括目标的三维位置信息(全球定位坐标系或大地坐标系中X、Y、Z坐标值)、目标的三维尺寸信息(长、宽、高)、目标的三维航向角信息(绕X-轴转动角度、绕Y-轴转动角度、绕Z-轴转动角度)、目标的三维速度信息(X-速度、Y-速度、Z-速度)、目标的三维加速度信息(X-加速度、Y-加速度、Z-加速度)、目标的三轴角速度信息(绕X-轴转动角速度、绕Y-轴转动角速度、绕Z-轴转动角速度)、目标的类别信息(轿车、行人、自行车、卡车、摩托车、未知等类别)、目标的置信度信息(超低、低、中、高、超高等等级)、目标的检测概率信息(一个大于等于0小于等于1的概率值)、目标的特殊行为信息(例如，目标行走看手机，目标开车看手机，目标开车打瞌睡等)等信息。The relevant information may include the three-dimensional position information of the target (X, Y, Z coordinate values in the global positioning coordinate system or the geodetic coordinate system), the three-dimensional size information of the target (length, width, height), and the three-dimensional heading angle information of the target (around the X-axis rotation angle, rotation angle around Y-axis, rotation angle around Z-axis), three-dimensional speed information of the target (X-speed, Y-speed, Z-speed), three-dimensional acceleration information of the target (X-acceleration, Y-acceleration, Z-acceleration), three-axis angular velocity information of the target (rotational angular velocity around X-axis, angular velocity around Y-axis, angular velocity around Z-axis), target category information (car, pedestrian, bicycle, truck) , motorcycle, unknown and other categories), the confidence information of the target (super-low, low, medium, high, super high and other grades), the detection probability information of the target (a probability value greater than or equal to 0 and less than or equal to 1), the special Behavioral information (for example, the target looks at the mobile phone while walking, the target looks at the mobile phone while driving, the target is dozing off while driving, etc.) and other information.
S20：根据相关信息依次计算每一目标的行为特征信息和属性特征信息。S20: Calculate the behavior feature information and attribute feature information of each target in turn according to the relevant information.
行为特征信息用于表征目标的行为特征，属性特征信息用于表征目标自身的特征。The behavioral feature information is used to represent the behavioral features of the target, and the attribute feature information is used to represent the characteristics of the target itself.
S30：根据行为特征信息和属性特征信息计算目标的整体重要度。S30: Calculate the overall importance of the target according to the behavior feature information and the attribute feature information.
S40：将整体重要度大于第一阈值的目标按照整体重要度的大小进行降序排列，以形成待选列表。S40: Arrange the targets whose overall importance is greater than the first threshold in descending order according to the overall importance to form a candidate list.
S50：在待选列表中从最前项依次往后选取不超过预设数量的目标，以形成重要目标列表。S50: In the to-be-selected list, select targets that do not exceed a preset number from the top item to the back to form an important target list.
本实施例的重要目标生成方法，根据目标的行为特征信息和属性特征信息分析其是否对本车具有重大影响，即计算整体重要度，并根据影响的大小(即整体重要度的大小)选出一定数量的重要目标，并形成降序排列的重要目标列表，以供车辆后续分析计算使用。本方法通过对车辆周边的目标进行细致的分析，获取到该重要目标列表。The method for generating an important target in this embodiment analyzes whether the target has a significant impact on the vehicle according to the behavior feature information and attribute feature information of the target, that is, calculates the overall importance, and selects a certain target according to the size of the impact (that is, the overall importance). The number of important targets, and a list of important targets in descending order is formed for the subsequent analysis and calculation of the vehicle. The method obtains the list of important targets by analyzing the targets around the vehicle in detail.
一个实施例中，行为特征信息包括：目标的行为异常度、行为鲁莽度和碰撞风险度。In one embodiment, the behavior feature information includes: abnormal behavior degree, behavior recklessness degree, and collision risk degree of the target.
另一个实施例中，属性特征信息包括：目标的自身脆弱度、重大威胁度和额外兴趣度。In another embodiment, the attribute feature information includes: the target's own vulnerability, major threat, and extra interest.
一个实施例中，S30中计算行为异常度包括以下步骤：In one embodiment, calculating the behavior abnormality degree in S30 includes the following steps:
获取目标当前位置的交通规则信息；Obtain the traffic rule information of the current location of the target;
获取目标当前的交通行为信息；Obtain the current traffic behavior information of the target;
根据交通规则信息和交通行为信息计算行为异常度。Behavior abnormality is calculated according to traffic rule information and traffic behavior information.
根据高精地图信息(车道线信息、地面标识信息、交通标识信息、交通信号灯信息等)和目标的定位信息(位置信息、姿态信息)，可以获取目标所处高精度地图中的位置和交通场景，以获取目标当前位置的交通规则信息，即在该位置应该按照何种交通规则行驶。根据目标的感知信息(类别、航向信息、速度信息、加速度信息、角速度信息等)，可以获取目标的具体的交通行为信息。然后根据高精地图中相关位置的交通规则信息和目标的具体交通行为信息，可以判断目标是否行为异常。According to the high-precision map information (lane line information, ground marking information, traffic sign information, traffic signal information, etc.) and the target's positioning information (position information, attitude information), the location and traffic scene of the target in the high-precision map can be obtained , to obtain the traffic rule information of the current location of the target, that is, what traffic rules should be followed in this location. According to the perception information of the target (category, heading information, speed information, acceleration information, angular velocity information, etc.), the specific traffic behavior information of the target can be obtained. Then, according to the traffic rule information of the relevant location in the high-precision map and the specific traffic behavior information of the target, it can be determined whether the target behaves abnormally.
一个实施例中，行为异常可以表现为三种形式，一种是目标的交通行为违反交通法规，极易引发交通事故，需要高度重点关注；另一种是其行为虽然没有违反交通规则但是由于目标位于潜在交通风险区域，有可能会导致潜在的交通事故，需要重点关注；还有一种是目标参与交通过程中注意力不集中，有可能做出危险的行为引发交通事故，也需要重点关注。In one embodiment, the abnormal behavior can be manifested in three forms. One is that the traffic behavior of the target violates traffic laws, which can easily lead to traffic accidents and requires high attention; the other is that although its behavior does not violate traffic rules, it is due to the It is located in a potential traffic risk area, which may lead to potential traffic accidents and needs to be paid attention to; another is that the target is inattentive when participating in the traffic process, and may make dangerous behaviors to cause traffic accidents, which also needs to be paid attention to.
第一类行为异常-目标违反交通规则：例如，在高速功能场景，如果车辆倒车行驶，则违反交规；在城市交通路口场景，如果行人在行人信号灯为红的状态下过马路，则违反交规。除此之外，还有很多方法可以判断目标是否违反交通法规。目标的第一类行为异常度ρabn1，可以通过对应目标所在位置、高精地图在目标所在位置的交通规则，以及目标的具体交通行为等信息计算得到。The first type of abnormal behavior - the target violates the traffic rules: For example, in the high-speed functional scene, if the vehicle reverses, it violates the traffic rules; in the urban traffic intersection scene, if the pedestrian crosses the road when the pedestrian signal is red, it violates the traffic rules. Beyond that, there are many ways to tell if a target is violating traffic laws. The abnormality degree ρabn1 of the first type of behavior of the target can be calculated by information such as the location of the target, the traffic rules of the high-precision map at the target location, and the specific traffic behavior of the target.
第二类行为异常-目标处于潜在交通风险区：例如，在高速公路场景，路肩上停靠一辆打着双闪的车辆，虽然这辆车没有违反交通法规，但是其他交通参与者通过时需要高度集中注意力，以防停靠车辆突然启动，或者从该车内或车后走出行人，带来突发交通事故。除此之外，还有很多方法可以判断目标是否行为异常会导致潜在的交通事故。目标的第二类行为异常度ρabn2，可以通过目标所在位置、目标的类别属性行为等特征、高精地图上目标所在位置的潜在风险等级等信息计算得到。The second category of abnormal behavior - the target is in a potential traffic risk area: for example, in a highway scene, a vehicle with double flashing stops on the shoulder of the road, although the vehicle does not violate traffic laws, but other traffic participants need to pass at a high altitude Pay attention to prevent a parked vehicle from starting suddenly, or walking out of a pedestrian from inside or behind the vehicle, causing an unexpected traffic accident. In addition to this, there are many ways to tell if a target is behaving abnormally that could lead to a potential traffic accident. The abnormality degree ρabn2 of the second type of behavior of the target can be calculated by information such as the location of the target, the characteristics of the target's category attribute and behavior, and the potential risk level of the target's location on the high-precision map.
第三类行为异常-目标注意力不集中：例如，行人边走边看书或手机，车辆驾驶员边开车边看手机等。除此之外，还有很多方法可以判断目标是否行为是否注意力不集中。目标的第三类行为异常度ρabn3，可以通过目标所在位置、目标的特殊行为信息、高精地图上目标所在位置特性等信息计算得到。The third type of abnormal behavior - the target's inattention: for example, pedestrians reading books or mobile phones while walking, vehicle drivers looking at mobile phones while driving, etc. In addition to this, there are many ways to determine whether the target is acting or not paying attention. The third type of behavioral anomaly degree ρabn3 of the target can be calculated from information such as the location of the target, the special behavior information of the target, and the location characteristics of the target on the high-precision map.
目标的行为异常度ρabn可以利用以上信息加权计算得到，其计算公式如下：The behavior abnormality degree ρabn of the target can be calculated by weighting the above information, and its calculation formula is as follows:
ρabn＝αabn1·ρabn1+αabn2·ρabn2+αabn3·ρabn3 (1)ρabn =αabn1 ·ρabn1 +αabn2 ·ρabn2 +αabn3 ·ρabn3 (1)
这里αabni为第i＝1,2,3,……类行为异常度的加权值(满足αabn1≥2αabn2≥6αabn3，而且αabn1+αabn2+αabn3＝1，意味着第一类行为异常度权重最高，第三类行为异常度权重最低，例如可以取αabn1＝0.6，αabn2＝0.3，αabn3＝0.1)，ρabni为前述计算得到的第i类行为异常度(满足0≤ρabni≤1，对于i＝1,2,3……)。这里目标的行为异常度满足0≤ρabn≤1。Here αabni is the weighted value of the i=1, 2, 3, ... category behavior abnormality (satisfies αabn1 ≥2α abn2 ≥ 6αabn3 , and αabn1 +αabn2 +αabn3 =1, which means that the first category The weight of the abnormality degree of behavior is the highest, and the weight of the degree of abnormal behavior of the third type is the lowest, for example, αabn1 =0.6, αabn2 =0.3, αabn3 =0.1), ρabni is the abnormal degree of the i-th type of behavior obtained by the above calculation (satisfying 0 ≤ρabni ≤1 for i=1,2,3...). Here, the behavioral anomaly degree of the target satisfies 0≤ρabn ≤1.
一个实施例中，S30中计算行为鲁莽度包括以下步骤：In one embodiment, calculating the behavior recklessness in S30 includes the following steps:
获取目标的运动信息变化率信息；Obtain the change rate information of the motion information of the target;
根据运动信息变化率信息计算行为鲁莽度。The behavioral recklessness is calculated based on the motion information change rate information.
目标行为鲁莽，指的是目标在参与交通的过程中表现出急躁、粗鲁、冲动、神经质、不受控制等比较极端的行为。产生这些鲁莽行为是多种多样的，一方面可能是由于目标(车辆)的操控者比较鲁莽导致被操控的目标车辆行为鲁莽，另一方面的原因可能是由于目标的制动、转向、发动机等装置出现故障导致目标失速、超速、翻转等看似鲁莽的行为。根据目标的感知信息(类别、航向信息、速度信息、加速度信息、角速度信息等)，可以判断目标是否行为鲁莽。Reckless behavior of the target refers to the extreme behavior of the target in the process of participating in traffic, such as impatience, rudeness, impulsiveness, neuroticism, and uncontrollability. There are many kinds of reckless behaviors. On the one hand, the operator of the target (vehicle) may be reckless, which leads to the reckless behavior of the manipulated target vehicle. On the other hand, the reason may be due to the braking, steering, engine, etc. of the target. A malfunction of the device results in the seemingly reckless behavior of the target stalling, overspeeding, flipping, etc. According to the perception information of the target (category, heading information, speed information, acceleration information, angular velocity information, etc.), it can be judged whether the target behaves recklessly.
目标的鲁莽度ρagg可以通过运动信息变化率信息(包括航向变化率、速度变化率、加速度变化率、加速度变化率等)计算得到，如果这些运动变化率至少一个以上的幅值超过设定的阈值，则认为目标行为鲁莽，越多的运动变化率超过阈值，则鲁莽度越高。这里的鲁莽度满足0≤ρagg≤1。The recklessness of the target ρagg can be calculated from the rate of change of the motion information (including the rate of change of the heading, the rate of speed, the rate of acceleration, the rate of acceleration, etc.) If the threshold is set, the target behavior is considered reckless, and the more the motion change rate exceeds the threshold, the higher the recklessness. The recklessness degree here satisfies 0≤ρagg ≤1.
一个实施例中，S30中计算碰撞风险度包括以下步骤：In one embodiment, calculating the collision risk in S30 includes the following steps:
计算目标与车辆发生碰撞的风险值；Calculate the risk value of the collision between the target and the vehicle;
根据风险值计算碰撞风险度。Calculate the collision risk based on the risk value.
根据目标的感知信息和自车的车载总线信息，可以判断本车是否会和目标发生潜在碰撞。目标和本车发生潜在碰撞的威胁越高，则其碰撞风险越高。例如，可以预测本车的行驶轨迹和目标的运动轨迹，如果两者轨迹相交并且到达相交点的时间接近，则可以判断本车和目标会发生潜在碰撞。According to the perception information of the target and the on-board bus information of the vehicle, it can be determined whether the vehicle will have a potential collision with the target. The higher the threat of a potential collision between the target and own vehicle, the higher the collision risk. For example, the driving trajectory of the vehicle and the motion trajectory of the target can be predicted. If the two trajectories intersect and the time to reach the intersection point is close, it can be determined that the vehicle and the target will have a potential collision.
目标的碰撞风险度ρcol可以通过自车速度信息、加速度信息、横摆角速度信息，目标的相对自车的位置信息、相对速度信息、相对加速度信息等获得。这里目标碰撞风险度满足0≤ρcol≤1。The collision risk degree ρcol of the target can be obtained from the ego vehicle speed information, acceleration information, yaw rate information, the relative ego vehicle position information, relative speed information, relative acceleration information, etc. of the target. Here, the target collision risk satisfies 0≤ρcol ≤1.
一个实施例中，S30中计算自身脆弱度包括以下步骤：In one embodiment, calculating the vulnerability of the self in S30 includes the following steps:
获取目标的分类信息、外观信息和尺寸信息；Obtain the classification information, appearance information and size information of the target;
根据分类信息、外观信息和尺寸信息计算自身脆弱度。Calculate own vulnerability according to classification information, appearance information and size information.
根据目标的分类、外观、体积等，可以判断目标是否比较脆弱，是否需要额外保护。例如，大部分非机动车辆(行人、自行车、驴车等)都是需要额外保护的目标，某些机动车辆也是需要额外保护的目标(两轮摩托车、三轮摩托车、微型薄皮轿车等)。根据目标的种类不同其自身脆弱程度不同，需要额外保护程度高的目标重要度也高(小孩和老人需要受保护的程度最高，成年人需要的受保护程度低，这是因为小孩和老人的行动力、判断力、反应速度弱于成年人)；目标的防护能力越弱，其自身脆弱程度越高(机动车辆的驾驶员外露的越多、外壳越薄、体积越小，则其越脆弱，在交通事故中越易受伤害)。According to the classification, appearance, volume, etc. of the target, it can be judged whether the target is relatively fragile and needs additional protection. For example, most non-motor vehicles (pedestrians, bicycles, donkey carts, etc.) are targets that require additional protection, and some motor vehicles are also targets that require additional protection (two-wheeled motorcycles, three-wheeled motorcycles, miniature cars, etc.) . Depending on the type of the target, its own vulnerability is different, and the target with a high degree of additional protection is also more important (children and the elderly need the highest degree of protection, and adults need the lowest degree of protection, because the behavior of children and the elderly is high. power, judgment, and reaction speed are weaker than adults); the weaker the target's protective ability, the higher its own vulnerability (the more the driver of the motor vehicle is exposed, the thinner the shell, and the smaller the volume, the more vulnerable it is. more vulnerable in a traffic accident).
目标的自身脆弱度ρfrg可以通过目标的分类信息、外观信息、尺寸信息等获得。这里目标的自身脆弱度满足0≤ρfrg≤1。The target's own vulnerability ρfrg can be obtained through the target's classification information, appearance information, size information and so on. Here, the target's own vulnerability satisfies 0≤ρfrg ≤1.
一个实施例中，S30中计算重大威胁度包括以下步骤：In one embodiment, calculating the major threat level in S30 includes the following steps:
获取目标的分类信息、外观信息、尺寸信息和外挂信息；Obtain the classification information, appearance information, size information and plug-in information of the target;
根据分类信息、外观信息、尺寸信息和外挂信息计算重大威胁度。Calculate the major threat level based on classification information, appearance information, size information and plug-in information.
根据目标的分类，可以判断目标是否是会产生重大威胁的交通参与者。例如，大型车辆(卡车、货车、巴士、工程车等)都是可能会产生重要大威胁的交通参与者。这是因为这些车辆的体积重量庞大，司机视角有限，车辆可能有外挂突出金属等物体，这些车辆一旦碰撞行人、小型车辆等会产生很大的伤害。According to the classification of the target, it can be judged whether the target is a traffic participant that poses a major threat. For example, large vehicles (trucks, vans, buses, construction vehicles, etc.) are all traffic actors that can pose a significant threat. This is because the volume and weight of these vehicles are huge, the driver's viewing angle is limited, and the vehicles may have objects such as protruding metal attached to the vehicle. Once these vehicles collide with pedestrians, small vehicles, etc., they will cause great damage.
目标的重大威胁度ρtht可以通过目标的分类信息、外观信息、尺寸信息、外挂信息等获得。这里目标的重大威胁度满足0≤ρtht≤1。The major threat degree ρtht of the target can be obtained through the classification information, appearance information, size information, plug-in information and so on of the target. Here, the major threat degree of the target satisfies 0≤ρtht ≤1.
一个实施例中，S30中计算额外兴趣度包括以下步骤：In one embodiment, calculating the additional interest degree in S30 includes the following steps:
接收第三方利益相关者发送的需要额外关注的目标列表；Receive a list of targets that require additional attention from third-party stakeholders;
根据目标列表计算额外兴趣度。Calculate additional interest from the target list.
通过无线通信，车辆可以实时接受到第三方利益相关者(例如，其他车辆、路基设施、交通管制部门、公共安全部门等)传输来得需要额外关注的目标列表。例如，交通管制部门发现一辆超速行驶车辆，其速度大大超过了速度限制，同时交警部门的监控系统和车辆没法及时追踪超速车辆，可以通过车路协同系统的无线通信，将超速车辆特征发布给系统内车辆和路基设施，通过别的交通参与者来检测超速车辆。同时也可以用于走失人员、犯罪分子追踪等。Through wireless communication, the vehicle can receive, in real time, a list of objects that require additional attention from third-party stakeholders (eg, other vehicles, road infrastructure, traffic control, public safety, etc.). For example, the traffic control department finds a speeding vehicle whose speed greatly exceeds the speed limit. At the same time, the monitoring system and the vehicle of the traffic police department cannot track the speeding vehicle in time. The characteristics of the speeding vehicle can be released through the wireless communication of the vehicle-road coordination system. For vehicles and roadbed facilities in the system, speeding vehicles are detected by other traffic participants. It can also be used to track lost people and criminals.
目标的额外兴趣度ρint可以通过目标的特征(外观、尺寸、标识、速度、车牌等特征)、第三方关注等级、赏金等级等计算得到。这里目标的额外兴趣度满足0≤ρint≤1。The additional interest degree ρint of the target can be calculated by the characteristics of the target (appearance, size, logo, speed, license plate, etc.), third-party attention level, bounty level, etc. Here the additional interest of the target satisfies 0≤ρint ≤1.
另一个实施例中，S30包括：In another embodiment, S30 includes:
对目标的行为异常度、行为鲁莽度、碰撞风险度、自身脆弱度、重大威胁度和额外兴趣度进行加权并取其中的最大值作为目标的整体重要度。The target's behavioral anomaly, behavioral recklessness, collision risk, self-vulnerability, major threat, and extra interest are weighted, and the maximum value is taken as the overall importance of the target.
根据前述目标的细分度量衡，通过如下方法计算得到目标的整体重要度ρimp，其计算公式如下：According to the aforementioned subdivision metrics of the target, the overall importance ρimp of the target is calculated by the following method, and the calculation formula is as follows:
ρimp＝max(ρcol，ρint，(αabn·ρabn+αagg·ρagg+αfrg·ρfrg)) (2)ρimp =max(ρcol , ρint , (αabn ·ρabn +αagg ·ρagg +αfrg ·ρfrg )) (2)
一个实施例中，碰撞风险度和额外兴趣度的加权系数均大于行为异常度、行为鲁莽度、自身脆弱度或重大威胁度的加权系数。即将碰撞风险度和额外兴趣度作为最重要的因素，行为异常度、行为鲁莽度、自身脆弱度和重大威胁度作为相对重要的因素。In one embodiment, the weighting coefficients of the collision risk degree and the extra interest degree are both larger than the weighting coefficients of the abnormal behavior degree, the behavior reckless degree, the self-vulnerability degree or the major threat degree. The collision risk and extra interest are the most important factors, and the abnormal behavior, the reckless behavior, the vulnerability and the major threat are relatively important factors.
这里的加权重要度的计算逻辑在于目标的碰撞风险度ρcol最重要，如果车辆将与目标发生碰撞产生重大安全事故，则该目标的将是最重要的目标，车辆需要重点关注，并做出相应的刹车、加速、转向等控制行为来避免碰撞发生。目标的额外兴趣度ρint也比较重要，这些额外感兴趣的目标有可能是对交通安全、公共安全的都比较重要的，要车辆重点关注以配合执法部门或相关机构进行观测。目标的异常度ρabn、鲁莽度ρagg、脆弱度ρfrg等都相对重要，这里通过加权的方式获取平均重要度，这里的权重0≤αabn≥1，0≤αagg≤1,0≤αfrg≤1而且αabn+αagg+αfrg＝1。最后，通过取最大值的方式，获得者几个重要度最大的那个作为目标的整体重要度。The calculation logic of the weighted importance here is that the collision risk ρcol of the target is the most important. If the vehicle will collide with the target and cause a major safety accident, the target will be the most important target, and the vehicle needs to focus on and make Corresponding control behaviors such as braking, acceleration, and steering are used to avoid collisions. The additional interest degree ρint of the target is also more important. These additional interest targets may be more important to traffic safety and public safety, and the vehicle should focus on it to cooperate with the law enforcement department or related agencies for observation. The abnormality ρabn , the recklessness ρagg , and the vulnerability ρfrg of the target are all relatively important. Here, the average importance is obtained by weighting, where the weights are 0≤αabn ≥1, 0≤αagg ≤1, 0≤ αfrg ≤ 1 and αabn + αagg + αfrg =1. Finally, by taking the maximum value, the one with the highest importance is obtained as the overall importance of the target.
当确定感兴趣目标的区域后，我们可以根据目标的特征、行为、轨迹来进一步判断目标的重要度，本发明提出了一种新的方法获取目标的重要度细分度量衡，最后通过加权平均的方式获得目标的重要度，最后输出重要度高的目标形成重要目标列表。After determining the area of the target of interest, we can further judge the importance of the target according to the characteristics, behavior and trajectory of the target. The present invention proposes a new method to obtain the subdivision measurement of the importance of the target. Finally, through the weighted average The importance of the target is obtained by the method, and finally the target with high importance is output to form an important target list.
至此，本领域技术人员应认识到，虽然本文已详尽示出和描述了本发明的多个示例性实施例，但是，在不脱离本发明精神和范围的情况下，仍可根据本发明公开的内容直接确定或推导出符合本发明原理的许多其他变型或修改。因此，本发明的范围应被理解和认定为覆盖了所有这些其他变型或修改。By now, those skilled in the art will recognize that, although various exemplary embodiments of the present invention have been illustrated and described in detail herein, the present invention may still be implemented in accordance with the present disclosure without departing from the spirit and scope of the present invention. The content directly determines or derives many other variations or modifications consistent with the principles of the invention. Accordingly, the scope of the present invention should be understood and deemed to cover all such other variations or modifications.
Claims (7)
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910745048.8ACN110349425B (en) | 2019-08-13 | 2019-08-13 | An important target generation method for vehicle-road cooperative autonomous driving system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910745048.8ACN110349425B (en) | 2019-08-13 | 2019-08-13 | An important target generation method for vehicle-road cooperative autonomous driving system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110349425A CN110349425A (en) | 2019-10-18 |
| CN110349425Btrue CN110349425B (en) | 2020-11-13 |
Family
ID=68184902
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910745048.8AActiveCN110349425B (en) | 2019-08-13 | 2019-08-13 | An important target generation method for vehicle-road cooperative autonomous driving system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110349425B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111599216A (en)* | 2020-05-18 | 2020-08-28 | 合肥杰代机电科技有限公司 | Auxiliary driving method, device and system based on image recognition and UWB (ultra-wideband) tag |
Family Cites Families (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9058247B2 (en)* | 2010-09-08 | 2015-06-16 | Toyota Jidosha Kabushiki Kaisha | Risk potential calculation apparatus |
| CN104773177A (en)* | 2014-01-09 | 2015-07-15 | 株式会社理光 | Aided driving method and aided driving device |
| JP2016148971A (en)* | 2015-02-12 | 2016-08-18 | トヨタ自動車株式会社 | Driving assistance device |
| JP6776513B2 (en)* | 2015-08-19 | 2020-10-28 | ソニー株式会社 | Vehicle control device, vehicle control method, information processing device, and traffic information provision system |
| EP3540712A4 (en)* | 2016-11-09 | 2019-11-20 | Sony Corporation | Information processing device, information processing method, program, and moving body |
| JP2018097590A (en)* | 2016-12-13 | 2018-06-21 | アイシン・エィ・ダブリュ株式会社 | Obstacle determination system and obstacle determination program |
| CN107316500B (en)* | 2017-06-06 | 2020-11-20 | 驭势（上海）汽车科技有限公司 | Threat degree calculation method in automatic driving, target selection method and application |
| CN108583571A (en)* | 2018-04-28 | 2018-09-28 | 深圳市商汤科技有限公司 | Collision control method and device, electronic equipment and storage medium |
- 2019
- 2019-08-13CNCN201910745048.8Apatent/CN110349425B/enactiveActive
Also Published As
| Publication number | Publication date |
|---|---|
| CN110349425A (en) | 2019-10-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113998034B (en) | Rider assistance system and method | |
| JP7499256B2 (en) | System and method for classifying driver behavior - Patents.com | |
| JP7638613B2 (en) | Information processing system, information processing method, and program | |
| CN107848533B (en) | Vehicle control device, vehicle control method, and medium for storing vehicle control program | |
| CN106255899B (en) | Device for signaling an object to a navigation module of a vehicle equipped with such a device | |
| CN113165652A (en) | Verifying predicted trajectories using a mesh-based approach | |
| US10336252B2 (en) | Long term driving danger prediction system | |
| CN110782657A (en) | Police cruiser using a subsystem of an autonomous vehicle | |
| CN113320541B (en) | Vehicle control device, vehicle control method, and storage medium | |
| CN108688598A (en) | Vehicle control system, control method for vehicle and the medium for storing vehicle control program | |
| US11104356B2 (en) | Display device and method for a vehicle | |
| CN111824135A (en) | driver assistance system | |
| JP2019535566A (en) | Unexpected impulse change collision detector | |
| US12051248B2 (en) | Moving body collision avoidance device, collision avoidance method and electronic device | |
| US11216001B2 (en) | System and method for outputting vehicle dynamic controls using deep neural networks | |
| KR102210140B1 (en) | Use of map information to smooth objects generated from sensor data | |
| US12277769B2 (en) | Method for deep neural network functional module deduplication | |
| CN110349425B (en) | An important target generation method for vehicle-road cooperative autonomous driving system | |
| WO2023132055A1 (en) | Evaluation device, evaluation method, and program | |
| CN108074395B (en) | A method and device for identification | |
| US12311876B2 (en) | Projected security zone | |
| CN120363931B (en) | Method and device for presenting assisted driving decision-making process | |
| JP7682405B2 (en) | Information gathering system and mobility support system | |
| CN115123140B (en) | Seat belt device for vehicle, tension control method, and storage medium | |
| US20250214579A1 (en) | Vehicle control device, vehicle control method, and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |