WO2018176593A1 - 一种面向无人自行车的局部避障路径规划方法 - Google Patents

一种面向无人自行车的局部避障路径规划方法 Download PDF

Info

Publication number
WO2018176593A1
WO2018176593A1 PCT/CN2017/084507 CN2017084507W WO2018176593A1 WO 2018176593 A1 WO2018176593 A1 WO 2018176593A1 CN 2017084507 W CN2017084507 W CN 2017084507W WO 2018176593 A1 WO2018176593 A1 WO 2018176593A1
Authority
WO
WIPO (PCT)
Prior art keywords
point
bicycle
obstacle
path
line
Prior art date
Application number
PCT/CN2017/084507
Other languages
English (en)
French (fr)
Inventor
吴建国
Original Assignee
深圳市靖洲科技有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 深圳市靖洲科技有限公司 filed Critical 深圳市靖洲科技有限公司
Publication of WO2018176593A1 publication Critical patent/WO2018176593A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G05CONTROLLING; REGULATING
    • G05DSYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
    • G05D1/00Control of position, course, altitude or attitude of land, water, air or space vehicles, e.g. using automatic pilots
    • G05D1/02Control of position or course in two dimensions
    • G05D1/021Control of position or course in two dimensions specially adapted to land vehicles
    • G05D1/0212Control of position or course in two dimensions specially adapted to land vehicles with means for defining a desired trajectory
    • G05D1/0223Control of position or course in two dimensions specially adapted to land vehicles with means for defining a desired trajectory involving speed control of the vehicle
    • GPHYSICS
    • G05CONTROLLING; REGULATING
    • G05DSYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
    • G05D1/00Control of position, course, altitude or attitude of land, water, air or space vehicles, e.g. using automatic pilots
    • G05D1/02Control of position or course in two dimensions
    • G05D1/021Control of position or course in two dimensions specially adapted to land vehicles
    • G05D1/0212Control of position or course in two dimensions specially adapted to land vehicles with means for defining a desired trajectory
    • G05D1/0214Control of position or course in two dimensions specially adapted to land vehicles with means for defining a desired trajectory in accordance with safety or protection criteria, e.g. avoiding hazardous areas

Definitions

  • the invention relates to an unmanned bicycle technology, in particular to a method for planning an obstacle avoidance path for an unmanned bicycle.
  • Baidu has announced the development of a complex artificial intelligence unmanned bicycle.
  • This product is an unmanned bicycle with complex artificial intelligence such as environmental awareness, planning and self-balancing control. It mainly integrates Baidu in artificial intelligence.
  • the achievements of deep learning, big data and cloud computing technologies however, there is no disclosure of technical details.
  • most of the sports intervention service systems with wide coverage, low cost and high specificity are adopted, and the intervention of the unmanned bicycles in accordance with the actual situation is expected to solve the problem of bicycle obstacle avoidance.
  • the obstacle avoidance path planning system determines how the vehicle reaches the target position under various constraints and path obstacle conditions, including environmental constraints embodied in safety, and systemic kinematic constraints embodying feasibility.
  • System dynamics constraints that reflect ride and stability, as well as specific optimization index constraints, such as the shortest time or shortest distance.
  • these constraints are concentrated in the global path planning.
  • the global path planning problem is equivalent to the problem of path generation between the starting point and the end point. Solving the global path planning problem generally requires a typical road to be learned in advance. And its digital storage method, that is, the environmental map, when the environmental changes or other factors lead to the planning results are not feasible, you need to restart the global plan to get a new feasible path to continue to exercise.
  • the main task of local path planning and trajectory generation is to ensure the safe and smooth form of bicycle. Firstly, the road information is obtained from the sensing system. After processing, a safe and smooth driving trajectory is generated in real time, and transmitted in the form of data of vehicle speed and steering angle. The control system enables the bicycle to achieve lane following and obstacle avoidance functions.
  • the so-called obstacle avoidance path planning refers to selecting a path from the starting point to the target point for a given obstacle condition and the initial and target poses, so that the moving object can pass all the obstacles safely and without collision.
  • the object of the present invention is to provide a local obstacle avoidance path planning method for an unmanned bicycle, comprising the following steps:
  • the environmental data obtained by using the environment-aware system will be unified into the same coordinate system, and an environmental map will be generated.
  • the environmental map includes two types: an obstacle map and a bicycle lane map;
  • the unmanned bicycle performs the obstacle avoidance step.
  • the obstacle map in the step (1) is given in a polar coordinate form, and the coordinate origin is a center point of the rear axle of the unmanned bicycle, and includes 720 data, that is, a data of 0.5 degrees, which is used to indicate the direction.
  • the closest object on the distance from the center of the bicycle, if there is no obstacle, its value is set to the maximum distance value.
  • the bicycle lane line map in the step (1) is composed of a group of lane line data, represented by two parts of lane line position data and line type data, and the position data is 10 sample points extracted on the lane line.
  • Cartesian coordinates, line type includes: roadside line, double yellow line, single solid line, single dotted line, parking line, zebra crossing, forbidden line, etc.
  • the coordinates of the center of the circle are O(x 0 , y 0 , z 0 ), and the coordinates of the reference point of the end effector of the unmanned bicycle are P(p x , p y , p z ).
  • the center of the obstacle is o
  • the radius is r
  • the starting point of the end effector of the unmanned bicycle is A
  • the target point is B.
  • the step (3) is performed according to the following process: (3-1) determining an unmanned bicycle working space, solving an obstacle center and an end effector space coordinate; (3-2) discretizing the space arc ADB Control point, determine the starting point and end point of the obstacle avoidance path; (3-3) Use the inverse kinematics solution to obtain the starting point, the end point and the discrete control point corresponding to the motion variables of each part of the unmanned bicycle; (3-4) Step (3)
  • the joint variables obtained in the middle are interpolated by cubic spline to obtain the motion function of each motion variable; (3-5) the primary driving path is generated; (3-6) the corresponding point of the motion function is taken, and the kinematics positive solution is used.
  • the principle of selecting the cubic spline interpolation point in the step (3-4) is: the first point: the center of the rear axle of the bicycle, and the second point: the first sampling point of the road center line and the first side of the right side line
  • the fourth point the tenth sampling point of the road center line and the tenth sampling point of the right sideline
  • the current side encounters a special key point when encountering an obstacle, and replaces some of the above four points by a translation process.
  • the step (3-5) generates a primary driving path, and adopts a simple and efficient arc curve as a driving path under special working conditions.
  • the curvature formula (3) and the steering angle formula ( 4) Generate a trajectory.
  • the step of generating a trajectory under the normal working condition is specifically: (3-5-1) 11 points are equally spaced on the path curve by the abscissa, and the path is divided into 10 segments; (3-5-2) t0, The t1 point is used as the starting point and the ending point of the first trajectory.
  • the position at time t0 is (0, 0)
  • the speed is the speed v at constant speed
  • the velocity component in the X and Y directions is (v, 0)
  • the position at t1 is derived from The path, and the maximum safe speed is obtained according to the turning radius of the path point.
  • v is taken as the speed of the time t1
  • the maximum safe speed is taken as the speed of the time t1
  • the X, Y-direction component is obtained from the slope of the point path. It is obtained that the distance between points t0 and t1 is divided by the speed average at two points of t1 and t2, and the result is taken as the value of t1; (3-5-3) the position and velocity at the point t1 and t1 are taken as four conditions.
  • Ten-segment trajectory constitutes a time-segment function, given by 0.005s Velocity at the time point and the steering angle value and transmitted to the bicycle chassis module.
  • the local path is updated in real time at a rate of 20 Hz to meet the needs of real-time generation of the trajectory.
  • the bicycle can be driven strictly according to the planned path, and the vehicle speed is automatically adjusted according to the path curvature, and in the case of moving or fixing the obstacle, the obstacle avoidance path planning can be performed in advance.
  • FIG. 1 is a schematic diagram of interpolation point selection according to an embodiment of the present invention.
  • FIG. 2 is a schematic diagram of an obstacle bounding box and an obstacle avoidance path according to an embodiment of the present invention
  • FIG. 3 is a schematic diagram of partial path planning of a special working condition according to an embodiment of the present invention.
  • FIG. 4 is a flow chart of a common frame path generation step according to an embodiment of the present invention.
  • the environmental data obtained by using the environment-aware system will be unified into the same coordinate system, and an environment map will be generated.
  • the environment map includes two types of obstacle maps and bicycle lane maps.
  • the obstacle map is given in the form of polar coordinates, and the origin of the coordinates is the center point of the rear axle of the unmanned bicycle, including 720 data, that is, a data of 0.5 degrees, which is used to indicate the linear distance of the nearest object in the direction from the center of the bicycle. If there is no obstacle, the value is set to the maximum distance value, and the bicycle lane line map is composed of a group of lane line data, which is represented by two parts of the lane line position data and the line type data, and the position data is
  • the rectangular coordinates of the 10 sampling points extracted on the lane line include: roadside line, double yellow line, single solid line, single dotted line, parking line, zebra crossing, and forbidden line.
  • the center line of the lane is used as the driving path, the equivalent bounding box is established, and the geometric center of the obstacle is determined.
  • the coordinates of the center of the circle are O(x 0 , y 0 , z 0 ), the reference point coordinates of the end effector of the unmanned bicycle are P(p x , p y , p z ), and when the formula (1) is established, the unmanned bicycle collides with the obstacle.
  • the center of the obstacle be o
  • the radius is r
  • the starting point of the end effector movement of the unmanned bicycle is A
  • the target point is B.
  • space curve ACDEB space curve AC'D'E'B
  • space curve AC"D"E"B space curve AC"D"E"B, etc., as shown in Figure 2, according to the shortest and end of the motion path
  • the space arc ADB is selected as the obstacle avoidance process path, wherein the D point is determined by the spatial size of the end effector and the obstacle and the safety factor of the collision avoidance;
  • the unmanned bicycle performs the following obstacle avoidance steps:
  • (3-1) Determine the working space of the unmanned bicycle, and solve the space coordinates of the obstacle center and the end effector;
  • step (3) (3-4) performing cubic spline interpolation on the joint variables obtained in step (3) to obtain a motion function of each motion variable;
  • the cubic spline curve has the advantage of the second derivative continuous on both sides of the interpolation point, the curvature of the cubic spline curve is continuous, which means that the bicycle steering will not be abrupt, and the bicycle can smooth the necessary conditions of the form, thereby enabling It is also possible for the driver to handle certain other matters while the bicycle is in motion.
  • the interpolation point selection diagram of the strength is selected, and the general selection principle of the key interpolation points when planning the path is:
  • the first point the center of the bicycle rear axle (ie the origin of the coordinates).
  • the second point the midpoint of the first sampling point of the road center line and the first sampling point of the right side line.
  • the third point the midpoint of the fifth sampling point of the road center line and the fifth sampling point of the right side line.
  • the fourth point the midpoint of the tenth sampling point of the road center line and the tenth sampling point of the right side line.
  • the current side encounters a special key point when encountering an obstacle, and replaces some of the above four points by a translation process.
  • the cubic spline curve is the driving path.
  • the slope of the ending point is according to the following formula. (2) Perform calculations:
  • the local path is updated in real time at a rate of 20 Hz to meet the needs of real-time generation of the trajectory.
  • curvature formula (3) is:
  • the t0, t1 point is taken as the starting point and the end point of the first trajectory, the position at time t0 is (0, 0), the speed is the speed v at constant speed, and the speed component in the X and Y directions is ( v, 0), the position at time t1 is derived from the path, and the maximum safe speed is obtained according to the turning radius of the path point.
  • v is taken as the speed of the time t1
  • the maximum safe speed is taken as the speed of the time t1
  • X The Y-direction component is obtained from the slope of the point path, and the distance between the points t0 and t1 is divided by the speed average at two points t1 and t2, and the result is taken as the value of t1;
  • the ten-segment trajectory constitutes a time-segment function, and the speed and steering angle values at various points in time are given at 0.005 s and transmitted to the bicycle chassis module.
  • (3-7) Use the joint motion function that meets the requirements to drive the joints of the human legs to achieve obstacle avoidance for unmanned bicycles.
  • the path of the design and the algorithm steps are used to obtain the trajectory of the end effector of the unmanned bicycle through the obstacle avoidance path simulation.
  • the motion speed is related to the interpolation speed.
  • the interpolation is uniform, the end effector is in the space trajectory and the planned path. It is consistent with the obstacle avoidance requirement of unmanned bicycle driving process in a certain speed range.
  • Its axial speed curve has monotonicity, the speed variable is within the constraint condition, the maximum speed of each axis satisfies the comprehensive characteristics of the unmanned bicycle, and the speed of each end of the end effector
  • the curve changes smoothly and continuously, which indicates that the endless actuator of the unmanned bicycle will not vibrate during the obstacle avoidance process, thus ensuring the smooth control and movement of the unmanned bicycle.

Landscapes

  • Engineering & Computer Science (AREA)
  • Aviation & Aerospace Engineering (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Remote Sensing (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Automation & Control Theory (AREA)
  • Traffic Control Systems (AREA)

Abstract

一种面向无人自行车的局部避障路径规划方法,可使得自行车严格按照规划路径行驶,并且车速自动根据路径曲率调整,遇到移动或者固定障碍物的情况下,可以提前进行避障路径规划,包括如下步骤:(1)利用环境感知系统所得环境数据统一到同一坐标系下,并生成环境地图,环境地图包含障碍物地图和自行车道线地图两种类型;(2)根据一般工况下自行车自主驾驶跟随自行车道线行驶的原则,采用拟合车道中心线作为行驶路径,建立当量包围盒,确定障碍物几何中心,确定避障过程路径;(3)根据无人自行车与障碍物发生碰撞的条件公式以及选取的路径,无人自行车执行避障步骤。

Description

一种面向无人自行车的局部避障路径规划方法 技术领域
本发明涉及无人自行车技术,特别是一种面向无人自行车避障路径规划方法。
背景技术
自20世纪60年代移动机器人诞生以来,研究人员一直梦想研究无人智能交通工具,作为智能交通系统的重要组成部分,无人自行车排除了人为不确定因素的影响,不仅可以提高驾驶安全性,而且可以解决交通拥堵,提高能源利用率,百度曾宣布开发复杂人工智能无人自行车,该产品是具备环境感知、规划和自平衡控制等复杂人工智能的无人自行车,主要集合了百度在人工智能、深度学习、大数据和云计算技术的成就,然而对技术细节没有任何披露。目前大多采用采用覆盖面广、成本低,且针对性强的运动干预服务系统,对无人自行车的运动进行符合实际情况的干预,有望解决自行车避障等问题。
作为无人自行车的智能核心,避障路径规划系统决定车辆如何在多种约束条件和路径障碍物条件下到达目标位置,这些约束包括体现为安全性的环境约束,体现可行性的系统运动学约束,体现平顺性和稳定性的系统动力学约束以及特定的优化指标约束,如最短时间或最短距离等。在无人自行车应用中,这些约束集中在全局路径规划中得到满足,全局路径规划问题等同于起点和终点间路径生成的问题,解决全局路径规划问题一般要求提前获知完成的典型道路 及其数字化存储方式,也就是环境地图,当环境变化或其他因素导致规划结果不可行时,需要重启全局规划得到新的可行路径才能继续行使。
然而目前无人自行车的工作环境与一般机器人应用存在很多不同,因此需要规划一种新的局部避障路径规划方法。局部路径规划及轨迹生成的主要任务是确保自行车安全、平顺的形式,其首先从感知系统获取道路信息,经过处理后实时生成安全、平顺的行车轨迹,并以车速和转向角的数据形式传输给控制系统,从而使自行车实现车道跟随和避障功能。
所谓避障路径规划是指在给定的障碍条件以及起始和目标的位姿,选择一条从起始点到达目标点的路径,使运动物体能安全、无碰撞的通过所有的障碍。
发明内容
本发明的目的在于提供一种面向无人自行车的局部避障路径规划方法,包括如下步骤:
(1)利用环境感知系统所得环境数据将统一到同一坐标系下,并生成环境地图,环境地图包含障碍物地图和自行车道线地图两种类型;
(2)根据一般工况下自行车自主驾驶跟随自行车道线行驶的原则,采用拟合车道中心线作为行驶路径,建立当量包围盒,确定障碍物几何中心,确定避障过程路径;
(3)根据无人自行车与障碍物发生碰撞的条件公式以及选取的路径,无人自行车执行避障步骤。
优选的,所述步骤(1)中的所述障碍物地图以极坐标形式给出,坐标原点为无人自行车后轴中心点,包括720个数据,即0.5度一个数据,用来表示该方向上的最近物体离自行车中心的直线距离,如果没有障碍物,其数值设为最大距离值。
优选的,所述步骤(1)中的所述自行车道线地图由一组车道线数据组成,用车道线位置数据和线型数据两部分表示,位置数据为车道线上所提取10个采样点的直角坐标,线型包含:路边线、双黄线、单实线、单虚线、停车线、斑马线、禁停线等。
优选的,所述步骤(2)中设圆心坐标为O(x0,y0,z0),无人自行车末端执行器参考点坐标为P(px,py,pz),当式(1)成立时,无人自行车与障碍物发生碰撞。
Figure PCTCN2017084507-appb-000001
优选的,所述步骤(2)中设障碍物圆心为o,半径为r,无人自行车末端执行器运动起始点为A,目标点为B,避障规划过程中,根据运动路径最短和末端执行器可达空间最大化原则选择多种曲线路径,其中末端点由末端执行器和障碍物的空间尺寸及避碰安全系数决定。
优选的,所述步骤(3)按照如下流程进行:(3-1)确定无人自行车工作空间,求解障碍物中心与末端执行器空间坐标;(3-2)将空间圆弧ADB离散化若干控制点,确定避障路径的起点与终点;(3-3)运用运动学逆解求得起点、终点与离散控制点对应无人自行车各部分运动变量;(3-4)对步骤(3)中所得关节变量进行三次样条插值,得到各运动变量的运动函数;(3-5)生成初级行车路径;(3-6)取运动函数极值对应点,进行运动学正解,利用式(1)判断是否与障碍物相碰,检验三次样条插值精度和离散点数目是否符合要求;(3-7)利用符合要求的关节运动函数驱动人的腿部关节,实现无人自行车避障。
优选的,所述步骤(3-4)中三次样条插值点的选取原则为:第一点:本自行车后轴中心,第二点:道路中心线第一采样点与右侧路边线第一采样点的中点,第三点:道路中心线第五采样点与右侧路边线第五采样点的中点,第四点:道路中心线第十采样点与右侧路边线第十采样点的中点,当前方遇到障碍物时产生特殊关键点,通过平移处理以取代上述四点中的部分关键点。
优选的,所述步骤(3)中根据关键点坐标插值生成三次样条曲线即为行车 路径,插值过程中使用第一边界条件,即始点和终点的斜率为给定值,始点的斜率取k0=0,终点的斜率根据如下公式(2)进行计算:
Figure PCTCN2017084507-appb-000002
式中,k1为终点处导数;x9为车道中心线第9点横坐标;x10为车道中心线第10点横坐标;y9为车道中心线第9点纵坐标;y10为车道中心线第10点纵坐标。
优选的,所述步骤(3-5)生成初级行车路径中,在特殊工况下采用简单高效的圆弧曲线作为行车路径,在普通工况下,根据曲率公式(3)和转向角公式(4)生成轨迹。
优选的,所述步骤(3-5)的曲率公式(3)为:
Figure PCTCN2017084507-appb-000003
式中,K表示曲线y(x)的斜率,根据转向运动学关系tanα=BK获得转向角的计算公式,即α=arctan(BK),式中B表示汽车轴距,α表示转向角。
所述普通工况下生成轨迹的步骤具体为:(3-5-1)在路径曲线上按横坐标等距选取11点,将路径分割成10段;(3-5-2)将t0,t1点作为第一段轨迹的起始点和终点,t0时刻的位置为(0,0),速度为匀速行驶时速度v,X,Y方向速度分量为(v,0),t1时刻位置来源于路径,并根据该路径点转弯半径求得最大安全车速,如果其大于v,则将v作为t1时刻车速,反之将最大安全车速作为t1时刻车速,其X,Y向分量由该点路径斜率求得,将t0,t1点间的距离除以t1,t2两点处速度均值,其结果作为t1的值;(3-5-3)以t0,t1点处的位置、速度作为4个条件联立方程求的第一段轨迹参数方程:X(t)=at3+bt2+ct+d,Y(t)=mt3+nt2+ot+p;(3-5-4)将t1时刻作为第二段轨迹的起始点按照(3-5-2)和(3-5-3)的规则循环,则可以分别求出第二至第十段轨迹参数方程;(3-5-5)十段轨迹组成时间分段函数,按0.005s给出各个时刻点处的速度和转向角数值并传送给自行车底盘模块。
局部路径以20Hz速率进行实时更新,以满足轨迹实时生成的需要。
采用本发明的避障局部路径规划方法,可使得自行车严格按照规划路径行驶,并且车速自动根据路径曲率调整,遇到移动或者固定障碍物的情况下,可以提前进行避障路径规划。
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。
附图说明
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。本发明的目标及特征考虑到如下结合附图的描述将更加明显,附图中:
图1为根据本发明实施例的插值点选择示意图;
图2为根据本发明实施例的障碍物包围盒以及避障路径示意图;
图3为根据本发明实施例的特殊工况局部路径规划示意图;
图4为根据本发明实施例的普通工框路径生成步骤流程图。
具体实施方式
结合附图如下详细说明该面向无人自行车的局部避障路径规划方法,包括如下步骤:
(1)利用环境感知系统所得环境数据将统一到同一坐标系下,并生成环境地图,环境地图包含障碍物地图和自行车道线地图两种类型。
其中,障碍物地图以极坐标形式给出,坐标原点为无人自行车后轴中心点,包括720个数据,即0.5度一个数据,用来表示该方向上的最近物体离自行车中心的直线距离,如果没有障碍物,其数值设为最大距离值,自行车道线地图由一组车道线数据组成,用车道线位置数据和线型数据两部分表示,位置数据为 车道线上所提取10个采样点的直角坐标,线型包含:路边线、双黄线、单实线、单虚线、停车线、斑马线、禁停线等。
(2)根据一般工况下自行车自主驾驶跟随自行车道线行驶的原则,采用拟合车道中心线作为行驶路径,建立当量包围盒,确定障碍物几何中心,设圆心坐标为O(x0,y0,z0),无人自行车末端执行器参考点坐标为P(px,py,pz),当式(1)成立时,无人自行车与障碍物发生碰撞。
Figure PCTCN2017084507-appb-000004
设障碍物圆心为o,半径为r,无人自行车末端执行器运动起始点为A,目标点为B。避障规划过程中,路径有多种选择:空间曲线ACDEB、空间曲线AC’D’E’B和空间曲线AC”D”E”B等,如图2所示,根据运动路径最短和末端执行器可达空间最大化原则,选取空间圆弧ADB作为避障过程路径,其中D点由末端执行器和障碍物的空间尺寸及避碰安全系数决定;
(3)根据式(1)以及选取的路径ADB,无人自行车执行如下避障步骤:
(3-1)确定无人自行车工作空间,求解障碍物中心与末端执行器空间坐标;
(3-2)将空间圆弧ADB离散化若干控制点,确定避障路径的起点与终点;
(3-3)运用运动学逆解求得起点、终点与离散控制点对应无人自行车各部分运动变量;
(3-4)对步骤(3)中所得关节变量进行三次样条插值,得到各运动变量的运动函数;
因为三次样条曲线具有在插值点两侧二阶导数连续的优点,因此三次样条曲线的曲率是连续的,也就意味着自行车转向不会发生突变,自行车可以平顺形式的必要条件,从而能够实现在自行车行驶过程中行车者还能处理一定的其他事务。
如图1所示,根据发明是实力的插值点选择示意图,而规划路径时关键插值点的一般选取原则为:
第一点:本自行车后轴中心(即坐标原点)。
第二点:道路中心线第一采样点与右侧路边线第一采样点的中点。
第三点:道路中心线第五采样点与右侧路边线第五采样点的中点。
第四点:道路中心线第十采样点与右侧路边线第十采样点的中点。
当前方遇到障碍物时产生特殊关键点,通过平移处理以取代上述四点中的部分关键点。
根据关键点坐标插值生成三次样条曲线即为行车路径,插值过程中使用第一边界条件,即始点和终点的斜率为给定值,始点的斜率取k0=0,终点的斜率根据如下公式(2)进行计算:
Figure PCTCN2017084507-appb-000005
式中k1为终点处导数;x9为车道中心线第9点横坐标;x10为车道中心线第10点横坐标;y9为车道中心线第9点纵坐标;y10为车道中心线第10点纵坐标。
(3-5)生成初级行车路径。
在特殊工况下,比如自行车转弯和掉头的时候采用简单高效的圆弧曲线作为行车路径,如图3所示。
局部路径以20Hz速率进行实时更新,以满足轨迹实时生成的需要。
在普通工况下,根据曲率公式(3)和转向角公式(4)按照附图4的步骤生成轨迹,其中曲率公式(3)为:
Figure PCTCN2017084507-appb-000006
式中,K表示曲线y(x)的斜率,根据转向运动学关系tanα=BK(5)获得转向角的计算公式(4),即α=arctan(BK),式中B表示汽车轴距,α表示转向角。
步骤具体为:
(3-5-1)在路径曲线上按横坐标等距选取11点,将路径分割成10段;
(3-5-2)将t0,t1点作为第一段轨迹的起始点和终点,t0时刻的位置为(0,0),速度为匀速行驶时速度v,X,Y方向速度分量为(v,0),t1时刻位置来源于路径,并根据该路径点转弯半径求得最大安全车速,如果其大于v,则将v作为t1时刻车速,反之将最大安全车速作为t1时刻车速,其X,Y向分量由该点路径斜率求得,将t0,t1点间的距离除以t1,t2两点处速度均值,其结果作为t1的值;
(3-5-3)以t0,t1点处的位置、速度作为4个条件联立方程求的第一段轨迹参数方程:
X(t)=at3+bt2+ct+d,Y(t)=mt3+nt2+ot+p
(3-5-4)将t1时刻作为第二段轨迹的起始点按照(3-5-2)和(3-5-3)的规则循环,则可以分别求出第二至第十段轨迹参数方程;
(3-5-5)十段轨迹组成时间分段函数,按0.005s给出各个时刻点处的速度和转向角数值并传送给自行车底盘模块。
(3-6)取运动函数极值对应点,进行运动学正解,利用式(1)判断是否与障碍物相碰,检验三次样条插值精度和离散点数目是否符合要求;
(3-7)利用符合要求的关节运动函数驱动人的腿部关节,实现无人自行车避障。
采用设计的路径以及算法步骤,通过避障路径仿真,得到无人自行车末端执行器轨迹,其运动速度与插补速度相关,当插补为匀速时,其末端执行器在空间轨迹与所规划路径吻合,符合一定速度范围内无人自行车行车过程避障要求,其各个轴向速度曲线具有单调性,速度变量在约束条件内,各轴最大速度满足无人自行车综合特性,末端执行器各轴速度曲线变化平滑、连续,表明无人自行车末端执行器在避障过程中不会发生振动,从而保证无人自行车的平稳控制与运动。
虽然本发明已经参考特定的说明性实施例进行了描述,但是不会受到这些实施例的限定而仅仅受到附加权利要求的限定。本领域技术人员应当理解可以在不偏离本发明的保护范围和精神的情况下对本发明的实施例能够进行改动和修改。

Claims (10)

  1. 一种面向无人自行车的局部避障路径规划方法,其特征在于包括如下步骤:
    (1)利用环境感知系统所得环境数据将统一到同一坐标系下,并生成环境地图,环境地图包含障碍物地图和自行车道线地图两种类型;
    (2)根据一般工况下自行车自主驾驶跟随自行车道线行驶的原则,采用拟合车道中心线作为行驶路径,建立当量包围盒,确定障碍物几何中心,确定避障过程路径;
    (3)根据无人自行车与障碍物发生碰撞的条件公式以及选取的路径,无人自行车执行避障步骤。
  2. 根据权利要求1所述的一种面向无人自行车的局部避障路径规划方法,其特征在于:所述步骤(1)中的所述障碍物地图以极坐标形式给出,坐标原点为无人自行车后轴中心点,包括720个数据,即0.5度一个数据,用来表示该方向上的最近物体离自行车中心的直线距离,如果没有障碍物,其数值设为最大距离值。
  3. 根据权利要求1所述的一种面向无人自行车的局部避障路径规划方法,其特征在于:所述步骤(1)中的所述自行车道线地图由一组车道线数据组成,用车道线位置数据和线型数据两部分表示,位置数据为车道线上所提取10个采样点的直角坐标,线型包含:路边线、双黄线、单实线、单虚线、停车线、斑马线、禁停线等。
  4. 根据权利要求1所述的一种面向无人自行车的局部避障路径规划方法,其特征在于:所述步骤(2)中设圆心坐标为O(x0,y0,z0),无人自行车末端执行器参考点坐标为P(px,py,pz),当式(1)成立时,无人自行车与障碍物发生碰撞。
    Figure PCTCN2017084507-appb-100001
  5. 根据权利要求4所述的一种面向无人自行车的局部避障路径规划方法,其特征在于:所述步骤(2)中设障碍物圆心为o,半径为r,无人自行车末端执行器运动起始点为A,目标点为B,避障规划过程中,根据运动路径最短和末端执行器可达空间最大化原则选择多种曲线路径,其中末端点由末端执行器和障碍物的空间尺寸及避碰安全系数决定。
  6. 根据权利要求1所述的一种面向无人自行车的局部避障路径规划方法,其特征在于:所述步骤(3)按照如下流程进行:(3-1)确定无人自行车工作空间,求解障碍物中心与末端执行器空间坐标;(3-2)将空间圆弧ADB离散化若干控制点,确定避障路径的起点与终点;(3-3)运用运动学逆解求得起点、终点与离散控制点对应无人自行车各部分运动变量;(3-4)对步骤(3)中所得关节变量进行三次样条插值,得到各运动变量的运动函数;(3-5)生成初级行车路径;(3-6)取运动函数极值对应点,进行运动学正解,利用式(1)判断是否与障碍物相碰,检验三次样条插值精度和离散点数目是否符合要求;(3-7)利用符合要求的关节运动函数驱动人的腿部关节,实现无人自行车避障。
  7. 根据权利要求6所述的一种面向无人自行车的局部避障路径规划方法,其特征在于:所述步骤(3-4)中三次样条插值点的选取原则为:第一点:本自行车后轴中心,第二点:道路中心线第一采样点与右侧路边线第一采样点的中点,第三点:道路中心线第五采样点与右侧路边线第五采样点的中点,第四点:道路中心线第十采样点与右侧路边线第十采样点的中点,当前方遇到障碍物时产生特殊关键点,通过平移处理以取代上述四点中的部分关键点。
  8. 根据权利要求1所述的一种面向无人自行车的局部避障路径规划方法,其特征在于:所述步骤(3)中根据关键点坐标插值生成三次样条曲线即为行车路径,插值过程中使用第一边界条件,即始点和终点的斜率为给定值,始点的斜率取k0=0,终点的斜率根据如下公式(2)进行计算:
    Figure PCTCN2017084507-appb-100002
    式中,k1为终点处导数;x9为车道中心线第9点横坐标;x10为车道中心线第10点横坐标; y9为车道中心线第9点纵坐标;y10为车道中心线第10点纵坐标。
  9. 根据权利要求6所述的一种面向无人自行车的局部避障路径规划方法,其特征在于:所述步骤(3-5)生成初级行车路径中,在特殊工况下采用简单高效的圆弧曲线作为行车路径,在普通工况下,根据曲率公式(3)和转向角公式(4)生成轨迹。
  10. 根据权利要求6所述的一种面向无人自行车的局部避障路径规划方法,其特征在于:所述步骤(3-5)的曲率公式(3)为:
    Figure PCTCN2017084507-appb-100003
    式中,K表示曲线y(x)的斜率,根据转向运动学关系tanα=BK获得转向角的计算公式,即α=arctan(BK),式中B表示汽车轴距,α表示转向角。
PCT/CN2017/084507 2017-03-31 2017-05-16 一种面向无人自行车的局部避障路径规划方法 WO2018176593A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201710210982.0A CN106873600A (zh) 2017-03-31 2017-03-31 一种面向无人自行车的局部避障路径规划方法
CN201710210982.0 2017-03-31

Publications (1)

Publication Number Publication Date
WO2018176593A1 true WO2018176593A1 (zh) 2018-10-04

Family

ID=59159912

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2017/084507 WO2018176593A1 (zh) 2017-03-31 2017-05-16 一种面向无人自行车的局部避障路径规划方法

Country Status (2)

Country Link
CN (1) CN106873600A (zh)
WO (1) WO2018176593A1 (zh)

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109508003A (zh) * 2018-12-05 2019-03-22 徐工集团工程机械股份有限公司 一种无人驾驶道路机械机群动态避让方法
CN112622934A (zh) * 2020-12-25 2021-04-09 上海商汤临港智能科技有限公司 一种参考轨迹点和参考轨迹生成方法、驾驶方法以及车辆
CN113205570A (zh) * 2021-04-30 2021-08-03 阿波罗智联(北京)科技有限公司 基于电子地图的电子斑马线的生成方法及装置
CN114287025A (zh) * 2019-06-14 2022-04-05 日产自动车株式会社 行驶辅助方法及行驶辅助装置
CN115489548A (zh) * 2022-09-20 2022-12-20 重庆大学 智能汽车园区道路路径规划方法
CN116263605A (zh) * 2023-04-21 2023-06-16 杭州国辰机器人科技有限公司 基于非线性优化的移动机器人平滑、实时碰撞规避方法
CN115489548B (zh) * 2022-09-20 2024-06-04 重庆大学 智能汽车园区道路路径规划方法

Families Citing this family (31)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107340772B (zh) * 2017-07-11 2020-05-22 北京超星未来科技有限公司 一种面向无人驾驶的拟人化参考轨迹规划方法
EP3494448A4 (en) 2017-07-13 2019-11-06 Beijing Didi Infinity Technology and Development Co., Ltd. SYSTEMS AND METHODS FOR TRACK DETERMINATION
CN107346611B (zh) * 2017-07-20 2021-03-23 北京纵目安驰智能科技有限公司 一种自主驾驶的车辆的避障方法以及避障系统
CN107856733B (zh) * 2017-11-07 2019-11-05 长春工业大学 一种面向人机和谐的汽车躲避动障碍物控制方法
CN108470159B (zh) * 2018-03-09 2019-12-20 腾讯科技(深圳)有限公司 车道线数据处理方法、装置、计算机设备和存储介质
CN108445886B (zh) * 2018-04-25 2021-09-21 北京联合大学 一种基于高斯方程的自动驾驶车辆换道规划方法及系统
CN110962856B (zh) * 2018-09-30 2021-07-27 毫末智行科技有限公司 用于确定车辆的环境目标所处的区域的方法及装置
CN109540159B (zh) * 2018-10-11 2020-11-27 同济大学 一种快速完备的自动驾驶轨迹规划方法
CN109471432B (zh) * 2018-11-08 2021-09-28 南京农业大学 一种自主导航农用车最短避障路径规划方法
CN109341699A (zh) * 2018-11-30 2019-02-15 四川九洲电器集团有限责任公司 一种基于避障转弯品质评估的智能无人平台路径规划方法
CN109855637A (zh) * 2018-12-24 2019-06-07 北京新能源汽车股份有限公司 一种车辆的自动驾驶路径规划方法、装置及设备
CN111796588A (zh) * 2019-03-22 2020-10-20 北京京东尚科信息技术有限公司 路径规划方法、路径规划装置及计算机可读存储介质
CN110083158B (zh) * 2019-04-28 2022-08-16 深兰科技(上海)有限公司 一种确定局部规划路径的方法和设备
CN111766891A (zh) * 2019-05-23 2020-10-13 北京京东尚科信息技术有限公司 用于控制无人机飞行的方法和装置
CN110287904B (zh) * 2019-06-27 2021-07-16 武汉中海庭数据技术有限公司 一种基于众包数据的车道线提取方法、装置及存储介质
CN110509923B (zh) * 2019-08-02 2020-10-16 重庆长安汽车股份有限公司 自动驾驶路径规划方法、系统、计算机可读存储介质及车辆
CN110758381B (zh) * 2019-09-18 2021-05-04 北京汽车集团有限公司 生成转向轨迹的方法、装置、存储介质及电子设备
CN112783144B (zh) * 2019-10-22 2023-09-29 舜宇光学(浙江)研究院有限公司 路径生成方法和路径规划方法及其系统和设备
CN111144064B (zh) * 2019-12-05 2022-07-19 稿定(厦门)科技有限公司 文字变形方法、介质、设备及装置
CN111444674B (zh) * 2020-03-09 2022-07-01 稿定(厦门)科技有限公司 文字变形方法、介质及计算机设备
CN113031583B (zh) * 2020-03-13 2022-11-25 青岛慧拓智能机器有限公司 一种结构化道路避障方法
CN111390909B (zh) * 2020-03-30 2022-06-28 上海应用技术大学 工业机器人的避障方法和系统
CN111459160B (zh) * 2020-04-02 2021-10-15 中国矿业大学 一种开放道路无人洗扫车大规模轨迹平滑方法
CN111552296B (zh) * 2020-05-14 2021-03-26 宁波智能装备研究院有限公司 一种基于弯曲柱坐标系的局部平滑轨迹规划方法
CN111693061A (zh) * 2020-06-18 2020-09-22 成都信息工程大学 一种动态路径规划中路径选择的方法
CN111665852B (zh) * 2020-06-30 2022-09-06 中国第一汽车股份有限公司 一种障碍物避让方法、装置、车辆及存储介质
CN112883909A (zh) * 2021-03-16 2021-06-01 东软睿驰汽车技术(沈阳)有限公司 基于包围盒的障碍物位置检测方法、装置和电子设备
CN113077643B (zh) * 2021-03-23 2022-05-03 上海振华重工(集团)股份有限公司 一种交通控制方法、交通控制系统及存储介质
CN113008261B (zh) * 2021-03-30 2023-02-28 上海商汤临港智能科技有限公司 一种导航方法、装置、电子设备及存储介质
CN113721625B (zh) * 2021-08-31 2023-07-18 平安科技(深圳)有限公司 Agv控制方法、装置、设备及存储介质
CN116383330B (zh) * 2023-06-06 2023-08-11 中航信移动科技有限公司 一种轨迹拟合方法、存储介质及电子设备

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7734387B1 (en) * 2006-03-31 2010-06-08 Rockwell Collins, Inc. Motion planner for unmanned ground vehicles traversing at high speeds in partially known environments
CN102353379A (zh) * 2011-07-06 2012-02-15 上海海事大学 一种适用于自动驾驶车导航的环境建模方法
CN102591332A (zh) * 2011-01-13 2012-07-18 同济大学 用于无人驾驶汽车局部路径规划的装置及方法
CN105759820A (zh) * 2016-04-08 2016-07-13 济宁中科先进技术研究院有限公司 基于激光和视觉的道路自主清扫控制系统及控制方法
CN105974917A (zh) * 2016-05-11 2016-09-28 江苏大学 一种基于新型人工势场法的车辆避障路径规划研究方法
CN106530380A (zh) * 2016-09-20 2017-03-22 长安大学 一种基于三维激光雷达的地面点云分割方法

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102902269B (zh) * 2012-09-21 2015-07-01 北京邮电大学 预选择最小距离指标冗余机器人动态避障方法
CN103995984A (zh) * 2014-06-09 2014-08-20 武汉科技大学 一种基于椭圆约束的机器人路径规划方法和装置
CN104407616B (zh) * 2014-12-03 2017-04-26 沈阳工业大学 基于免疫网络算法的移动机器人动态路径规划方法
US10386844B2 (en) * 2015-09-30 2019-08-20 Deere & Company System and method for using geo-fenced guidance lines

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7734387B1 (en) * 2006-03-31 2010-06-08 Rockwell Collins, Inc. Motion planner for unmanned ground vehicles traversing at high speeds in partially known environments
CN102591332A (zh) * 2011-01-13 2012-07-18 同济大学 用于无人驾驶汽车局部路径规划的装置及方法
CN102353379A (zh) * 2011-07-06 2012-02-15 上海海事大学 一种适用于自动驾驶车导航的环境建模方法
CN105759820A (zh) * 2016-04-08 2016-07-13 济宁中科先进技术研究院有限公司 基于激光和视觉的道路自主清扫控制系统及控制方法
CN105974917A (zh) * 2016-05-11 2016-09-28 江苏大学 一种基于新型人工势场法的车辆避障路径规划研究方法
CN106530380A (zh) * 2016-09-20 2017-03-22 长安大学 一种基于三维激光雷达的地面点云分割方法

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109508003A (zh) * 2018-12-05 2019-03-22 徐工集团工程机械股份有限公司 一种无人驾驶道路机械机群动态避让方法
CN114287025A (zh) * 2019-06-14 2022-04-05 日产自动车株式会社 行驶辅助方法及行驶辅助装置
CN112622934A (zh) * 2020-12-25 2021-04-09 上海商汤临港智能科技有限公司 一种参考轨迹点和参考轨迹生成方法、驾驶方法以及车辆
CN113205570A (zh) * 2021-04-30 2021-08-03 阿波罗智联(北京)科技有限公司 基于电子地图的电子斑马线的生成方法及装置
CN113205570B (zh) * 2021-04-30 2024-04-19 阿波罗智联(北京)科技有限公司 基于电子地图的电子斑马线的生成方法及装置
CN115489548A (zh) * 2022-09-20 2022-12-20 重庆大学 智能汽车园区道路路径规划方法
CN115489548B (zh) * 2022-09-20 2024-06-04 重庆大学 智能汽车园区道路路径规划方法
CN116263605A (zh) * 2023-04-21 2023-06-16 杭州国辰机器人科技有限公司 基于非线性优化的移动机器人平滑、实时碰撞规避方法

Also Published As

Publication number Publication date
CN106873600A (zh) 2017-06-20

Similar Documents

Publication Publication Date Title
WO2018176593A1 (zh) 一种面向无人自行车的局部避障路径规划方法
Sharma et al. Recent advances in motion and behavior planning techniques for software architecture of autonomous vehicles: A state-of-the-art survey
Fan et al. Baidu apollo em motion planner
CN106371439B (zh) 一种统一的自动驾驶横向规划方法与系统
Khodayari et al. A historical review on lateral and longitudinal control of autonomous vehicle motions
WO2019042295A1 (zh) 一种无人驾驶路径规划方法、系统和装置
Hubschneider et al. Integrating end-to-end learned steering into probabilistic autonomous driving
US11577723B2 (en) Object trajectory association and tracking
US20230192127A1 (en) Autonomous vehicle trajectory generation using velocity-based steering limits
Sharma et al. A survey on smooth path generation techniques for nonholonomic autonomous vehicle systems
Tang et al. An overview of path planning algorithms
Chen et al. Target-driven obstacle avoidance algorithm based on DDPG for connected autonomous vehicles
Yu et al. Hierarchical framework integrating rapidly-exploring random tree with deep reinforcement learning for autonomous vehicle
Zhang et al. An intelligent driver model with trajectory planning
Fu et al. Collision-free and kinematically feasible path planning along a reference path for autonomous vehicle
Nan et al. Interaction-Aware Planning With Deep Inverse Reinforcement Learning for Human-Like Autonomous Driving in Merge Scenarios
Huang et al. Trajectory planning in frenet frame via multi-objective optimization
Wang et al. A hybrid trajectory planning strategy for intelligent vehicles in on-road dynamic scenarios
Wang et al. Vision-Based Autonomous Driving: A Hierarchical Reinforcement Learning Approach
Huang et al. General Optimal Trajectory Planning: Enabling Autonomous Vehicles with the Principle of Least Action
Sun et al. Get-dipp: Graph-embedded transformer for differentiable integrated prediction and planning
Narula et al. Two-level hierarchical planning in a known semi-structured environment
Zeng et al. A steerable curvature approach for efficient executable path planning for on-road autonomous vehicle
Zhang et al. Intelligent path planning by an improved RRT algorithm with dual grid map
Li et al. Decision making for autonomous vehicles

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 17902658

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 20/01/2020)

122 Ep: pct application non-entry in european phase

Ref document number: 17902658

Country of ref document: EP

Kind code of ref document: A1