Method, apparatus and computer program product for increasing positioning precision of vehicle

1 . A method for increasing a positioning precision of a vehicle, comprising:

obtaining an actually measured position of the vehicle;

detecting a vehicle environment feature corresponding to the actually measured position;

computing a position compensation value based on the vehicle environment feature; and

using the position compensation value to compensate the actually measured position to increase the positioning precision of the vehicle,

wherein detecting the vehicle environment feature comprises:

obtaining a time corresponding to the actually measured position;

obtaining an image relating to a surrounding environment of the vehicle captured by a sensor of the vehicle at the time;

identifying a blocking object in the image;

detecting an object feature of the blocking object, the object feature comprising at least one of (i) a category of the blocking object, (ii) a shape of the blocking object, (iii) a size of the blocking object, and (iv) a position of the blocking object relative to the vehicle; and

combining the object feature and/or the time to form the vehicle environment feature.

2 . The method according to claim 1 , wherein identifying the blocking object in the image comprises:

identifying an identified object having a predefined category from the image using image recognition technology, the predefined category comprising at least one of multi-story buildings, trees, bridges, and tunnels; and

the identified object is the blocking object.

3 . The method according to claim 1 , wherein computing the position compensation value comprises:

computing the position compensation value based on the vehicle environment feature using a compensation value computing model.

4 . An apparatus for increasing the positioning precision of the vehicle, comprising:

a processor; and

a non-transitory memory, storing a computer-executable instruction which, when executed, causes the processor to implement the method according to claim 1 .

5 . A computer program product for increasing the positioning precision of the vehicle, comprising a computer program configured to be executed by a processor to implement the method according to claim 1 .

6 . A method for increasing a positioning precision of a vehicle, comprising:

obtaining an actually measured position of the vehicle;

detecting a vehicle environment feature corresponding to the actually measured position;

computing a position compensation value based on the vehicle environment feature using a compensation value computing model; and

using the position compensation value to compensate the actually measured position to increase the positioning precision of the vehicle,

wherein training of the compensation value computing model comprises (i) generating a training data set using a test vehicle and a position simulation system, and (ii) using the training data set to train the compensation value computing model.

7 . The method according to claim 6 , wherein generating the training data set comprises:

obtaining a set of actually measured position samples;

obtaining a set of simulated position samples outputted by the position simulation system;

forming multiple error position pairs using the set of actually measured position samples and the set of simulated position samples;

generating multiple training samples based on the multiple error position pairs; and

combining the multiple training samples to form the training data set.

8 . The method according to claim 7 , wherein the set of actually measured position samples and the set of simulated position samples are generated under identical conditions, the identical conditions comprising identical vehicles, identical starting points, identical end points and identical routes of travel.

9 . The method according to claim 7 , wherein forming multiple error position pairs comprises, for each actually measured position sample in the set of actually measured position samples:

extracting, from the set of simulated position samples, a simulated position sample having a time equal to the time of the actually measured position sample;

computing a difference value between the actually measured position sample and the simulated position sample;

determining whether the difference value exceeds a predetermined threshold; and

in response to determining that the difference value exceeds the predetermined threshold, combining the actually measured position sample and the simulated position sample to form an error position pair.

10 . The method according to claim 7 , wherein generating multiple training samples comprises, for each of the multiple error position pairs:

detecting a vehicle environment feature sample corresponding to an actually measured position sample in the error position pair;

generating a position compensation value label corresponding to the vehicle environment feature sample by computing a difference value between the actually measured position sample and a simulated position sample in the error position pair; and

combining the vehicle environment feature sample and the position compensation value label to form a training sample corresponding to the error position pair.

11 . The method according to claim 10 , wherein detecting the vehicle environment feature sample comprises:

obtaining a time sample corresponding to the error position pair;

obtaining an image sample relating to a surrounding environment of the test vehicle captured by a sensor of the test vehicle at said time sample;

identifying a blocking object sample in the image sample;

detecting an object feature sample of the blocking object sample, the object feature sample comprising at least one of (i) a category of the blocking object sample, (ii) a shape of the blocking object sample, (iii) a size of the blocking object sample, and (iv) a position of the blocking object sample relative to the test vehicle; and

combining the object feature sample and/or the time sample to form the vehicle environment feature sample.

12 . An apparatus for increasing a positioning precision of a vehicle, comprising:

a processor configured to:

obtain an actually measured position of the vehicle;

detect a vehicle environment feature corresponding to the actually measured position;

compute a position compensation value based on the vehicle environment feature using a compensation value computing model; and

use the position compensation value to compensate the actually measured position to increase the positioning precision of the vehicle,

wherein detecting the vehicle environment feature comprises:

obtaining a time corresponding to the actually measured position;

obtaining an image relating to a surrounding environment of the vehicle captured by a sensor of the vehicle at the time;

identifying a blocking object in the image;

detecting an object feature of the blocking object, the object feature comprising at least one of (i) a category of the blocking object, (ii) a shape of the blocking object, (iii) a size of the blocking object, and (iv) a position of the blocking object relative to the vehicle; and

combining the object feature and/or the time to form the vehicle environment feature.

13 . The apparatus according to claim 12 , wherein training of the compensation value computing model comprises:

generating a training data set using a test vehicle and a position simulation system; and

using the training data set to train the compensation value computing model.