Warning method of obstacles and device of obstacles

1. An obstacle warning method for an obstacle avoidance and path planning system of a vehicle, at least comprising steps of:

acquiring a first scenario image at a current sampling time, a second scenario image at a previous sampling time, and a map about first relative distances between respective viewing points in a viewing field and the vehicle;

dividing and marking an obstacle in the first scenario image in accordance with the map about the first relative distances so as to acquire a profile and marking information of the obstacle, and weighted-averaging first distances in the profile so as to acquire a map about a second relative distance between the obstacle and the vehicle;

calculating a motion vector of the obstacle from the current sampling time to the previous sampling time in accordance with the first scenario image and the second scenario image;

matching the marked obstacle acquired at the current sampling time with the corresponding obstacle acquired at the previous sampling time in accordance with the profile, the marking information and the motion vector, and calculating a map about a third relative distance between the obstacle and the vehicle at the previous sampling time in accordance with the profile of the obstacle at the previous sampling time and the map about the first relative distances between the respective viewing points and the vehicle at the previous sampling time;

calculating a change value of a relative distance between the obstacle and the vehicle at the current sampling time and the previous sampling time in accordance with the map about the second relative distance and the map about the third relative distance, and calculating a speed of the obstacle relative to the vehicle at the current sampling time in accordance with the change value of the relative distance and an interval between the current sampling time and the previous sampling time; and

determining a collision time when the obstacle acquired at the current sampling time is to collide with the vehicle in accordance with the map about the second relative distance, the profile of the obstacle and the speed of the obstacle relative to the vehicle, and sending an obstacle warning in accordance with the collision time.

2. The obstacle warning method according to claim 1 , wherein the step of acquiring the first scenario image at the current sampling time, the second scenario image at the previous sampling time, and the map about the first relative distances between respective viewing points in the viewing field and the vehicle comprises:

based on a binocular distance-measurement principle, acquiring, by a binocular camera device, the first scenario image, the second scenario image and the map about the first relative distances between the respective viewing points in the viewing field and the vehicle in accordance with a vision disparity between two scenario images of the obstacle generated at an identical time as well as a structural parameter of the binocular camera device; or

based on a monocular distance-measurement principle, performing, by a monocular camera device, geometrical conversion on an image at the current sampling time and an image at the previous sampling time, so as to acquire the first scenario image and the second scenario image at the current sampling, and the map about the first relative distances between the respective viewing points in the viewing field and the vehicle; or

based on a radar distance-measurement principle, scanning, by a radar scanning device, the obstacle in a radar scanning range, so as to acquire the first scenario image and the second scenario image at the current sampling time and the previous sampling time, and the map about the first relative distances between the respective viewing points in the viewing field and the vehicle.

3. The obstacle warning method according to claim 1 , wherein the step of acquiring the map about the second relative distance between the obstacle and the vehicle at the current sampling time comprises: based on the monocular distance-measurement principle, training an imaging function of the camera on the basis of a template, so as to mark a feature point or identify a target through machine learning; directly extracting a profile of the identified target; and marking the profile and a size of the stored template, so as to directly acquire the map about the second relative distance between the obstacle and the vehicle at the current sampling time.

4. The obstacle warning method according to claim 1 , wherein the step of dividing and marking the obstacle in the first scenario image in accordance with the map about the first relative distances so as to acquire the profile and marking information of the obstacle, and weighted-averaging the first distances in the profile so as to acquire the map about the second relative distance between the obstacle and the vehicle comprises:

filtering out the map about the first relative distances at the current sampling time, so as to remove a road surface and a false obstacle;

acquiring a second relative distance between the obstacle and the vehicle at the current sampling time in accordance with the map about the first relative distances, and connecting the viewing points, distances between which and the vehicle are equal to the second relative distance, to form the profile; and

marking the profile, and weighted-averaging the first relative distances in the profile to mark the second relative distance between the obstacle and the vehicle.

5. The obstacle warning method according to claim 1 , wherein the step of calculating the motion vector of the obstacle from the current sampling time to the previous sampling time comprises:

dividing each of the first scenario image at the current sampling time and the second scenario image at the previous sampling time into N blocks, N being a positive integer;

calculating a correspondence between each block in the first scenario image at the current sampling time and the corresponding block in the second scenario image at the previous sampling time using a quick block-matching algorithm or a frequency-domain algorithm, so as to acquire a motion vector field of each block, or calculating a correspondence between a critical point in the first scenario image at the current sampling time and a corresponding critical point in the second scenario image at the previous sampling time using an optical flow algorithm or the frequency-domain algorithm, so as to acquire a motion vector field of the critical point; and

assigning or interpolating the motion vector field, so as to calculate the motion vector of the obstacle from the current sampling time to the previous sampling time.

6. The obstacle warning method according to claim 1 , wherein the step of calculating the speed of the obstacle relative to the vehicle at the current sampling time in accordance with the change value of the relative distance and the interval between the current sampling time and the previous sampling time comprises:

dividing the change value of the second relative distance by the interval between the current sampling time and the previous sampling time, so as to acquire the speed of the obstacle relative to the vehicle at the current sampling time.

7. The obstacle warning method according to claim 1 , wherein the step of determining the collision time when the obstacle acquired at the current sampling time is to collide with the vehicle in accordance with the map about the second relative distance, the profile of the obstacle and the speed of the obstacle relative to the vehicle comprises:

weighted-averaging second distances in the profile in accordance with the map about the second relative distance between the obstacle and the vehicle at the current sampling time and the profile of the obstacle, so as to acquire a second relative distance between the obstacle and the vehicle at the current sampling time; and

dividing the second relative distance between the obstacle and the vehicle at the current sampling time by the speed of the obstacle relative to the vehicle at the current sampling time, so as to acquire the collision time when the obstacle acquired at the current sampling time is to collide with the vehicle.

8. The obstacle warning method according to claim 1 , further comprising:

determining a position of the obstacle and a relative distance between the obstacle and the vehicle at a next sampling time in accordance with the map about the second relative distance between the obstacle and the vehicle at the current sampling time, the profile of the obstacle, the speed of the obstacle relative to the vehicle, and the motion vector of the obstacle from the current sampling time to the previous sampling time.

9. The obstacle warning method according to claim 7 , wherein the step of determining the position of the obstacle and the relative distance between the obstacle and the vehicle at the next sampling time comprises:

weighted-averaging motion vectors of the obstacle from the current sampling time to the previous sampling time, so as to acquire the motion vector of the obstacle from the current sampling time to the previous sampling time;

determining the position of the obstacle at the next sampling time in accordance with the motion vector of the obstacle from the current sampling time to the previous sampling time; and

determining the relative distance between the obstacle and the vehicle at the next sampling time in accordance with the map about the second relative distance between the obstacle and the vehicle at the current sampling time, the speed of the obstacle relative to the vehicle, and the interval between the current sampling time and the previous sampling time.