KR20230056225A - Obstacle avoidance and path tracking method considering the kinetic dynamics of a differential driving robot - Google Patents

Abstract

An obstacle avoidance and path tracking method considering the kinematics of a differentially driven mobile robot comprises: a first step of setting a non-obstacle area with the radius being the shortest distance between a robot and a surrounding obstacle; a second step of determining the intersection between a pre-given global path and the non-obstacle area as an non-obstacle target position; a third step of generating a non-obstacle target speed for moving to the non-obstacle target position by considering the allowable range between the linear and angular speeds of a robot; a fourth step of generating a speed window area considering the current speed, maximum allowable speed, and maximum allowable acceleration of the robot; a fifth step of selecting a window target speed considering the speed window area and the non-obstacle target speed; a sixth step of calculating the angular range of an obstacle extension area resulting from the overlap between an area where the obstacle is extended by the radius of the robot (obstacle extension area) and the non-obstacle area; a seventh step of selecting the final control speed to be transmitted to a robot controller by selecting the window target speed as the final control speed when the window target speed of the fifth step does not belong to an obstacle collision speed area where a collision with an obstacle is expected in the angular velocity and velocity coordinate system of the sixth step; and an eighth step of performing local planning by executing a path generation algorithm within a local map to generate a local path with the intersection between the border of the local map and the global path as the target point and the position of the robot as the starting point when the non-obstacle target position of the second step exists within the obstacle extension area or the intersection between the global path and the non-obstacle area does not exist (when the non-obstacle target position is not defined). Accordingly, it is possible to efficiently follow a given global path without colliding with an obstacle.

Description

Obstacle avoidance and path tracking method considering the kinetic dynamics of a differential driving robot

The present invention relates to a method for estimating a path of a mobile robot, and more particularly, to a method for avoiding obstacles and following a path considering kinematics of a differential drive type mobile robot.

A general path planning algorithm or a path planning algorithm for minimizing the amount of computation does not consider the kinematics of the robot and creates a path assuming that movement in any direction is possible (holonomic constraint).

Therefore, even if the robot controller is given a global path without obstacle collision, it is difficult to follow an error-free path due to kinematic limitations of the robot in the process of following it, and in this process, it may collide with obstacles on the map.

In addition, dynamic obstacles that are not on the map may be located on the global path during movement and may interfere with following the global path.

KR 10-1784500 B

The present invention has been proposed to solve the above technical problems, and efficiently provides a given global path without collision with an obstacle by considering the kinematics of a differential drive type mobile robot for a global path generated without considering the constraints of the robot. An obstacle avoidance and path following method considering the kinematics of a differential drive type mobile robot that generates a followable speed command is provided.

According to an embodiment of the present invention for solving the above problem, in the obstacle avoidance and path following method considering the kinematics of a differential drive type mobile robot, an obstacle-free area is set with a radius of the shortest distance between the robot and surrounding obstacles. A first step of doing, a second step of determining the intersection of a given global path and an obstacle-free area as an obstacle-free target position, and a step for moving to an obstacle-free target position considering the allowable range between the linear velocity and the angular velocity of the robot. The third step of generating the target speed without obstacles, the fourth step of generating a speed window area considering the current speed, the maximum allowable speed and the maximum allowable acceleration of the robot, and the window target considering the speed window area and the target speed without obstacles The fifth step of selecting the speed, the sixth step of calculating the angular range of the obstacle extension region generated by the overlap between the obstacle extension region (obstacle extension region) and the non-obstacle region extending the obstacle by the radius of the robot, and the fifth step If the window target speed does not belong to the obstacle collision speed range in which a collision with an obstacle is expected in the angular speed and speed coordinate system of step 6, the window target speed is selected as the final control speed and the final control speed transmitted to the robot controller is selected. Step 2, if the non-obstacle target location of the second step exists within the obstacle extension area or if there is no intersection between the global path and the non-obstacle area (when the non-obstacle target location is not defined), a path generation algorithm within the local map An eighth step of executing local planning to generate a local path with the intersection point between the edge of the local map and the global path as a target point and the location of the robot as a starting point is provided.

In addition, in the first step in the present invention, the obstacle-free area centered on the robot measures the distance to obstacles around the robot based on the LIDAR distance sensor, and among them, the maximum distance between the robot and the nearest obstacle is determined as a radius. It is characterized by being defined as a circle.

Further, in the present invention, in the seventh step, when the window target speed of the fifth step belongs to the obstacle collision speed region of the sixth step, the area belonging to the difference between the speed window region of the fourth step and the obstacle extension region of the sixth step It is characterized in that the angular velocity and the linear velocity pair closest to the window target velocity are used as the final control velocity and transmitted to the robot controller.

An obstacle avoidance and path following method according to an embodiment of the present invention can efficiently follow a given global path without colliding with an obstacle by considering the kinematics of a differential drive type mobile robot for the generated global path without considering the constraints of the robot. It is an algorithm that generates speed commands that can be The speed command generated by the corresponding algorithm is transmitted to the differential drive type robot controller to move the robot by controlling its wheels.

That is, the proposed obstacle avoidance and path following method can be used as an obstacle avoidance algorithm during autonomous driving of a differential drive type mobile robot. For example, when an automated guided vehicle (AGV) is operated in a smart factory, it can move safely by avoiding collisions with workers or other automated guided vehicles (AGVs).

1 is a diagram showing the progress of an obstacle avoidance and path following method according to an embodiment of the present invention

Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings in order to describe in detail enough for those skilled in the art to easily implement the technical idea of the present invention.

The inputs of the proposed algorithm (obstacle avoidance and path following method) are six types of information ((1) the global path generated by the preset path planning algorithm based on the global map, (2) the current speed of the robot, and (3) the robot’s The maximum allowable speed, (4) the maximum allowable acceleration of the robot, (5) the current position of the robot, (6) the data of the distance measuring sensor (LIDAR, etc.) mounted on the robot), and this algorithm (obstacle avoidance and path following method) The output of is two control speed commands ((1) linear velocity, (2) angular velocity) transmitted to the robot controller.

Existing algorithms used for the same purpose as the proposed algorithm (obstacle avoidance and path following method) include dynamic window approach (DWA) based on sampling and timed elastic band (TEB), which performs optimization by transforming a given global path. , vector field histogram-based algorithms (VFH+, VFH*), etc.

In the case of DWA, the performance is greatly affected by the number of samples, and it is necessary to tune the parameters to calculate the cost function for each sample. In the case of TEB, a graph optimization algorithm for nonlinear optimization must be applied, and in this process, a large amount of computation and parameter tuning are required.

In the case of the VFH-based algorithm, the robot's minimum rotation radius is limited to a value greater than 0, so it cannot generate commands for stationary rotation, and also does not consider the current speed, maximum speed, and maximum acceleration of the robot.

The proposed algorithm (obstacle avoidance and path following method) can generate a speed command that can efficiently avoid obstacles while considering the kinematics of a differential drive mobile robot with a low computational complexity compared to the existing algorithms.

1 is a diagram showing the progress of an obstacle avoidance and path following method according to an embodiment of the present invention.

Referring to FIG. 1, the proposed algorithm (obstacle avoidance and path following method) repeatedly performs the following 8-step process.

1. Set the obstacle-free area with the radius of the shortest distance between the robot and surrounding obstacles

The obstacle-free area centered on the robot measures the distance to obstacles around the robot based on a distance sensor such as LIDAR, and is a circle whose radius is the maximum distance between the robot and the nearest obstacle.

Since there is no obstacle inside the obstacle-free area, the amount of calculation can be minimized in the process of determining whether there is a collision between the robot and the obstacle.

2. Determine the intersection of the pre-given global path and the non-obstacle area as the non-obstacle target location

It is assumed that the global route is generated in advance through a preset route generation algorithm before autonomous driving and is provided to the proposed algorithm.

Since there are no obstacles in the obstacle-free area, there are no obstacles on the global path connecting the robot and the obstacle-free target position.

Therefore, the robot moves to the non-obstacle target position, which continuously changes during the movement process, as a destination.

3. Generate the target velocity for moving to the target position without obstacles considering the allowable range between the linear velocity and the angular velocity of the differential drive robot

Differential drive robots can drive in curves while moving forward, as well as turn in place.

Therefore, it is necessary to generate a smooth target velocity capable of curve and rotation in place by considering the distance between the forward direction of the robot and the target position, the angle, and the constraint relationship between linear velocity and angular velocity.

However, it is assumed that the robot does not move backward. That is, the linear velocity is greater than or equal to zero.

For this, g(q,q _t ) is defined.

The linear velocity V _T (d, θ) is calculated as follows based on equation (1).

는 각각 크루징 속도, 목표지점 도달 시의 목표속도, 목표지점과의 거리, 목표지점 도달 시의 속도로 변경하기 위한 임계거리, 타겟과 로봇간의 각도, 속도 조절을 위한 임계각도 이다.here

are the cruising speed, the target speed when reaching the target point, the distance from the target point, the critical distance for changing to the speed when reaching the target point, the angle between the target and the robot, and the critical angle for speed control.

However, d means the distance to the final destination, not the non-obstacle target position, while θ is the angle with the non-obstacle target position.

In Equation (2), v(d) is a function that starts converting the speed of V _cruise into _Vgoal when the distance between the robot and the final destination is less than d _t and finally arrives at the final destination at the speed of V _goal . .

s(θ) is a scaler function that becomes 0 when the angle between the non-obstacle target position and the forward direction of the robot is greater than θ ^v _t , making the linear velocity 0, and converging to a value of 1 as the angle difference decreases. Through this, if the angular difference is greater than the critical angle, the linear speed is controlled so that the robot is directed to the non-obstacle target position with a short turning radius and angular speed, and the linear speed is continuously increased as the angular difference decreases.

- The angular velocity w _T (θ) is calculated as:

Here, θ ^w _b is a critical angle, and when the magnitude of θ is greater than the critical angle, the value obtained by multiplying the sign of θ by W _cruise is used as the angular velocity, and the robot rotates toward the non-obstacle target position at a fast angular velocity, and when θ is 0 As it gets closer, the angular velocity converges to 0, so that the forward direction of the robot is fixed in the direction of the non-obstacle target position.

In order to satisfy the constraints between linear velocity and angular velocity, the following inequality must be satisfied.

In addition, by setting the critical angle to be smaller than π/2 in Equation (4), when the magnitude of the angular difference between the forward direction of the robot and the direction of the non-obstacle target position is greater than π/2, the linear velocity becomes 0, and the angular velocity It has a cruising angular velocity and rotates in place.

4. Creation of the speed window area considering the robot's current speed, maximum allowable speed, and maximum allowable acceleration

Same as the method of the Dynamic Window Approach (DWA), the current angular velocity and linear velocity of the robot are centered in the angular velocity-linear velocity coordinate system, and the maximum allowable speed of the robot, the maximum allowable acceleration, the control cycle time, and the line of the differential drive type mobile robot A velocity window area is created considering the allowable range of velocity and angular velocity.

By selecting an acceleration and velocity pair that exists within the velocity window region, the constraints on the robot's velocity and acceleration can be considered.

5. Window target speed selection considering the speed window area and non-obstacle target speed

If the obstacle-free target velocity obtained in the third step is within the velocity window region of the fourth step, the corresponding angular velocity and linear velocity pair is selected as the window target velocity.

Otherwise, the speed closest to the non-obstacle target speed within the speed window area is selected as the window target speed.

6. Calculation of the angular range of the obstacle expansion area caused by the overlap between the obstacle extension area (obstacle expansion area) and the obstacle-free area

Using a distance sensor such as LIDAR, the angle range of the obstacle is calculated by measuring the start angle and end angle of the space where the non-obstacle area and the obstacle extension area overlap.

Therefore, in the obstacle-free area, the robot's movement path that meets the arc (obstacle arc) existing between the start angle and the end angle of the obstacle expansion area is expected to collide with the obstacle, and this movement path is determined by the robot's speed command.

If the starting and ending angles of the obstacle are α and β, respectively, and the inequality of -π/2 < α< β < π/2 is established, the collision velocity region of the obstacle in the angular velocity and velocity coordinate system is expected to be as follows.

사이의 공간이 장애물 충돌 속도영역이 된다.Here, R ₀ is the radius of the non-obstacle area, and when |α|<ε (ε is a small value close to 0), ω=0, that is, the v-axis and

The space between them becomes the obstacle collision speed region.

The reason why the range of the obstacle start angle and end angle is -π/2 to π/2 is that when the angle difference between the robot and the target in Equation (4) is greater than or equal to π/2, s(θ) in Equation (2) is Since it becomes 0, it rotates in place, and in this case, collision between the robot and the obstacle does not occur. Obstacle start and end angles outside the range of -π/2 to π/2 are ignored.

7. Final control speed selection

If the window target speed of step 5 does not belong to the obstacle collision speed range of step 6, the window target speed is selected as the final control speed and transmitted to the robot controller.

If the window target velocity of step 5 belongs to the obstacle collision velocity region of step 6, the angular velocity and linear velocity pair closest to the window target velocity within the region belonging to the difference set between the region of step 4 and the region of step 6 is used as the final control speed and transmitted to the robot controller.

8. Do local planning

If the obstacle-free target location of the second step exists within the obstacle expansion area or if there is no intersection between the global path and the obstacle-free area (ie, when the obstacle-free target location is not defined), the path generation algorithm in the local map Execute to create a local path with the intersection point between the border of the local map and the global path as the target point and the robot's location as the starting point.

The local map is a cost map (COSTMAP) with a relatively smaller size than the global map, centered on the robot and continuously updated from distance sensor information.

Since the route planning algorithm is performed on the local map, the amount of computation is less compared to using the global map, so real-time route generation is possible.

An obstacle avoidance and path following method according to an embodiment of the present invention can efficiently follow a given global path without colliding with an obstacle by considering the kinematics of a differential drive type mobile robot for the generated global path without considering the constraints of the robot. It is an algorithm that generates speed commands that can be The speed command generated by the corresponding algorithm is transmitted to the differential drive type robot controller to move the robot by controlling its wheels.

That is, the proposed obstacle avoidance and path following method can be used as an obstacle avoidance algorithm during autonomous driving of a differential drive type mobile robot. For example, when an automated guided vehicle (AGV) is operated in a smart factory, it can move safely by avoiding collisions with workers or other automated guided vehicles (AGVs).

As such, those skilled in the art to which the present invention pertains will be able to understand that the present invention may be embodied in other specific forms without changing its technical spirit or essential features. Therefore, the embodiments described above should be understood as illustrative in all respects and not limiting. The scope of the present invention is indicated by the following claims rather than the detailed description above, and all changes or modifications derived from the meaning and scope of the claims and their equivalent concepts should be interpreted as being included in the scope of the present invention. do.

Claims (3)

In the obstacle avoidance and path following method considering the kinematics of a differential drive type mobile robot,
A first step of setting an obstacle-free area with a radius of the shortest distance between the robot and surrounding obstacles;
a second step of determining an intersection between a pre-given global path and an obstacle-free area as an obstacle-free target position;
A third step of generating an obstacle-free target velocity for moving to a target-free position in consideration of an allowable range between the linear velocity and the angular velocity of the robot;
A fourth step of generating a speed window area considering the current speed, maximum allowable speed, and maximum allowable acceleration of the robot;
a fifth step of selecting a window target velocity in consideration of the velocity window area and the non-obstacle target velocity;
a sixth step of calculating an angular range of an extended obstacle area generated by overlapping an area obtained by extending an obstacle by the radius of the robot (an obstacle extended area) and a non-obstacle area;
If the window target speed in step 5 does not belong to the obstacle collision speed range in which a collision with an obstacle is expected in the angular velocity and speed coordinate system in step 6, the final control speed transmitted to the robot controller by selecting the window target speed as the final control speed a seventh step of selecting; and
When the non-obstacle target location of the second step exists within the obstacle extension area or there is no intersection between the global path and the non-obstacle area (when the non-obstacle target location is not defined), the path creation algorithm is executed in the local map. An eighth step of performing local planning to generate a local path using the intersection of the edge of the local map and the global path as a target point and the location of the robot as a starting point;
Obstacle avoidance and path following method comprising a.
According to claim 1,
The first step is
The obstacle-free area centered on the robot measures the distance to obstacles around the robot based on the lidar distance sensor, and is defined as a circle with the maximum distance between the robot and the closest obstacle among them as a radius. Evasion and path following methods.
According to claim 1,
Step 7 is
If the window target velocity of step 5 belongs to the obstacle collision speed region of step 6, the angular velocity closest to the window target velocity within the region belonging to the difference between the velocity window region of step 4 and the obstacle extension region of step 6 An obstacle avoidance and path following method characterized in that the linear speed pair is used as the final control speed and transmitted to the robot controller.

Images