Terminal, method and device for recognizing obstacle, and storage medium

1. A terminal configured to recognize an obstacle, comprising: a motion sensor, an ultrasonic sensor, and a processor; wherein:

the motion sensor and the ultrasonic sensor are electrically coupled to the processor;

the motion sensor is configured to acquire a first dynamic parameter of the terminal;

the ultrasonic sensor is configured to acquire a second dynamic parameter of the terminal; and

the processor is configured to:

perform motion analysis on the obstacle based on the first dynamic parameter and the second dynamic parameter, to obtain an event recognition result of the obstacle;

input the first dynamic parameter into a first obstacle recognition model, to obtain a first velocity recognition result of at least one velocity recognition result;

input the second dynamic parameter into a second obstacle recognition model, to obtain a second velocity recognition result of the at least one velocity recognition result and a first distance recognition result of at least one distance recognition result;

determine an event recognition result of a previous moment as a first recognition result;

obtain a second recognition result of the obstacle based on the first velocity recognition result and the second velocity recognition result;

determine the second recognition result as the event recognition result when the first velocity recognition result and the second velocity recognition result satisfy a first preset condition; and

determine the first recognition result as the event recognition result when the first velocity recognition result and the second velocity recognition result do not satisfy the first preset condition.

2. The terminal according to claim 1 , wherein the first velocity recognition result comprises any one of an approaching event, a departing event, or an uncertain event, the second velocity recognition result comprises any one of the approaching event, the departing event, or the uncertain event;

the processor is configured to:

determine the second recognition result as the approaching event when the first velocity recognition result and the second velocity recognition result comprise the approaching event and do not comprise the departing event;

determine the second recognition result as the departing event when the first velocity recognition result and the second velocity recognition result comprise the departing event and do not comprise the approaching event; and

determine the second recognition result as the uncertain event when the first velocity recognition result and the second velocity recognition result comprise the departing event and the approaching event, or the first velocity recognition result and the second velocity recognition result are both uncertain events.

3. The terminal according to claim 1 , wherein the first distance recognition result comprises any one of an approaching event, a departing event, or an uncertain event, the second recognition result comprises any one of the approaching event, the departing event, or the uncertain event;

the first preset condition comprises: the first distance recognition result comprising the approaching event and not comprising the departing event when the second recognition result being the approaching event; and the first distance recognition result comprising the departing event and not comprising the approaching event when the second recognition result being the departing event.

4. The terminal according to claim 1 , wherein the processor is configured to:

input the first dynamic parameter and the second dynamic parameter into a third obstacle recognition model, to obtain the event recognition result;

wherein the third obstacle recognition model is configured to recognize the obstacle, and is obtained by training a neural network model using a historical first dynamic parameter and a historical second dynamic parameter as samples.

5. The terminal according to claim 4 , wherein the processor is configured to:

extract n first velocity vectors from the first dynamic parameter;

extract n second velocity vectors and n first distance vectors from the second dynamic parameter;

combine the i th first velocity vector, the i th second velocity vector and the i th first distance vector, to obtain n first vector sets; and

input the n first vector sets into the third obstacle recognition model sequentially, to obtain the event recognition result, where i and n are positive integers, and i≤n.

6. The terminal according to claim 4 , wherein the third obstacle recognition model comprises:

a three-layer fully connected neural network and an output layer connected in sequence; or

a multi-recurrent neural network core, a one-layer fully connected neural network and the output layer connected in sequence.

7. The terminal according to claim 1 , wherein the terminal further comprises a touch screen coupled to the processor;

the processor is configured to:

acquire a third dynamic parameter of the terminal at a current moment by the touch screen;

input the first dynamic parameter, the second dynamic parameter and the third dynamic parameter into a corresponding obstacle recognition model, to obtain the at least one velocity recognition result and the at least one distance recognition result; and

determine the event recognition result based on the at least one velocity recognition result and the at least one distance recognition result.

8. The terminal according to claim 7 , wherein the processor is configured to:

input the first dynamic parameter into the first obstacle recognition model, to obtain the first velocity recognition result;

input the second dynamic parameter into the second obstacle recognition model, to obtain the second velocity recognition result and the first distance recognition result; and

input the third dynamic parameter into a fourth obstacle recognition model, to obtain a second distance recognition result of the at least one distance recognition result.

9. The terminal according to claim 8 , wherein the processor is configured to:

determine the event recognition result of a previous moment as the first recognition result;

obtain the second recognition result of the obstacle based on the first velocity recognition result and the second velocity recognition result;

determine the second recognition result as the event recognition result when the first velocity recognition result and the second velocity recognition result satisfy a second preset condition; and

determine the first recognition result as the event recognition result when the first velocity recognition result and the second velocity recognition result do not satisfy the second preset condition.

10. The terminal according to claim 9 , wherein the first velocity recognition result comprises any one of an approaching event, a departing event, or an uncertain event, the second velocity recognition result comprises any one of the approaching event, the departing event, or the uncertain event;

the processor is configured to:

determine the second recognition result as the approaching event when the first velocity recognition result and the second velocity recognition result comprise the approaching event and do not comprise the departing event;

determine the second recognition result as the departing event when the first velocity recognition result and the second velocity recognition result comprise the departing event and do not comprise the approaching event;

determine the second recognition result as the uncertain event when the first velocity recognition result and the second velocity recognition result comprise the departing event and the approaching event, or the first velocity recognition result and the second velocity recognition result are both uncertain events.

11. The terminal according to claim 9 , wherein the first distance recognition result comprises any one of an approaching event, a departing event, or an uncertain event, the second distance recognition result comprises any one of the approaching event, the departing event, or the uncertain event, the second recognition result comprises any one of the approaching event, the departing event, or the uncertain event;

the second preset condition comprises: the first distance recognition result and the second distance recognition result comprising the approaching event and not comprising the departing event when the second recognition result being the approaching event; and the first distance recognition result and the second distance recognition result comprising the departing event and not comprising the approaching event when the second recognition result being the departing event.

12. The terminal according to claim 7 , wherein the seventh obstacle recognition model comprises at least one of:

a three-layer fully connected neural network and an output layer connected in sequence; or

a multi-recurrent neural network core, a one-layer fully connected neural network and the output layer connected in sequence;

wherein the terminal is a mobile phone configured to realize recognition of the obstacle collectively with the motion sensor and the ultrasonic sensor, correct deviation caused by noise over the ultrasonic sensor, thereby improving obstacle recognition accuracy.

13. The terminal according to claim 1 , wherein the at least one velocity recognition result further comprises a third velocity recognition result, the at least one distance recognition result further comprises a third distance recognition result;

the processor is configured to:

input the first dynamic parameter and the second dynamic parameter into a fifth obstacle recognition model to obtain the third velocity recognition result; and

input the second dynamic parameter and a third dynamic parameter into a sixth obstacle recognition model to obtain the third distance recognition result.

14. The terminal according to claim 13 , wherein the third velocity recognition result comprises any one of an approaching event, a departing event, or an uncertain event, the third distance recognition result comprises any one of the approaching event, the departing event, or the uncertain event;

the processor is configured to:

determine the event recognition result as the approaching event when the third velocity recognition result and the third distance recognition result comprise the approaching event and do not comprise the departing event;

determine the event recognition result as the departing event when the third velocity recognition result and the third distance recognition result comprise the departing event and do not comprise the approaching event; and

determine the event recognition result as the uncertain event when the third velocity recognition result and the third distance recognition result comprise the approaching event and the departing event, or the third velocity recognition result and the third distance recognition result are both uncertain events.

15. The terminal according to claim 1 , wherein the terminal further comprises a touch screen coupled to the processor;

the processor is configured to:

acquire a third dynamic parameter of the terminal at a current moment by the touch screen; and

input the first dynamic parameter, the second dynamic parameter and the third dynamic parameter into a seventh obstacle recognition model, to obtain the event recognition result, wherein the seventh obstacle recognition model is configured to recognize the obstacle, and is obtained by training a neural network model using a historical first dynamic parameter, a historical second dynamic parameter and a historical third dynamic parameter as samples.

16. The terminal according to claim 15 , wherein the processor is configured to:

extract n first velocity vectors from the first dynamic parameter;

extract n second velocity vectors and n first distance vectors from the second dynamic parameter;

extract n second distance vectors from the third dynamic parameter;

combine the i th first velocity vector, the i th second velocity vector, the i th first distance vector and the i th second distance vector, to obtain n vector sets; and

input the n vector sets into the seventh obstacle recognition model sequentially, to obtain the event recognition result, where i and n are positive integers, and i≤n.

17. A method for recognizing an obstacle, comprising:

acquiring a first dynamic parameter of a terminal at a current moment;

acquiring a second dynamic parameter of the terminal at the current moment;

performing motion analysis on the obstacle based on the first dynamic parameter and the second dynamic parameter, to obtain an event recognition result of the obstacle;

inputting the first dynamic parameter into a first obstacle recognition model, to obtain a first velocity recognition result of at least one velocity recognition result;

inputting the second dynamic parameter into a second obstacle recognition model, to obtain a second velocity recognition result of the at least one velocity recognition result and a first distance recognition result of at least one distance recognition result;

determining an event recognition result of a previous moment as a first recognition result;

obtaining a second recognition result of the obstacle based on the first velocity recognition result and the second velocity recognition result;

determining the second recognition result as the event recognition result when the first velocity recognition result and the second velocity recognition result satisfy a first preset condition; and

determining the first recognition result as the event recognition result when the first velocity recognition result and the second velocity recognition result do not satisfy the first preset condition.