Robot cleaner and operating method thereof

1. A robot cleaner configured to avoid a stuck situation through an artificial intelligence (AI), the robot cleaner comprising:

at least one sensor configured to acquire sensing information;

a driving motor configured to drive the robot cleaner;

a memory configured to store a compensation model to infer escape path factors from a learning surrounding map; and

at least one processor configured to:

acquire a surrounding map, based on the sensing; information,

determine escape path factors based on the surrounding map by using the compensation model, when the stuck situation of the robot cleaner is detected, and

control the driving motor so that the robot cleaner travels, based on the determined escape path factors,

wherein the at least one processor is configured to:

assign a reward or a penalty to the compensation model, based on at least one of whether the robot cleaner succeeds in escaping from the stuck situation or an escape time,

assign a reward of a first value to the compensation model, when the robot cleaner succeeds in escaping from the stuck situation within a preset time, and

assign a reward of a second value less than the first value to the compensation model when the robot cleaner succeeds in escaping from the stuck situation, after the preset time.

2. The robot cleaner of claim 1 , wherein the escape path factors include at least one of a rotation angle, a rotation direction, a forward direction, or a backward direction.

3. The robot cleaner of claim 2 , wherein the at least one sensor includes:

a bumper sensor configured to sense a bumper event; and

a depth sensor configured to acquire three-dimensional (3D) data, and

wherein the surrounding map is generated by the bumper event and the 3D data.

4. The robot cleaner of claim 3 , wherein the compensation model is a model based on an artificial neural network which is trained through reinforcement learning in a deep learning algorithm or a machine learning algorithm, and

wherein the reinforcement learning is performed through a Markov decision process.

5. The robot cleaner of claim 4 , wherein the at least one processor is configured to:

acquire the escape path factors based on an output result of the compensation model from the learning surrounding model, and

acquire, based on the acquired escape path factors, whether the robot cleaner succeeds in escaping from the stuck situation and the escape time taken until the robot cleaner escapes the stuck situation.

6. The robot cleaner of claim 3 , wherein the bumper sensor measures an impulse applied to a bumper provided in the robot cleaner, and generates the bumper event when the impulse is equal to or greater than a preset impulse, and

wherein the at least one processor is configured to determine the robot cleaner as being in the stuck situation, when number of times that the bumper event occurs is equal to or greater than a preset number of times.

7. The robot cleaner of claim 6 , wherein a cliff sensor is provided at a lower surface of the bumper of the robot cleaner, and

wherein the cliff sensor is configured to measure a distance between the floor and the cliff sensor using a transmitted infrared signal and a reflected infrared signal.

8. The robot cleaner of claim 7 , wherein the at least one processor is configured to determine that the robot cleaner reaches a staircase or a cliff when the measured distance is equal to or greater than a certain distance or when the reflected infrared signal is not detected for a certain time.

9. The robot cleaner of claim 3 , wherein the surrounding map includes at least one obstacle.

10. The robot cleaner of claim 2 , wherein the driving motor includes:

a left wheel motor to drive a left wheel; and

a right wheel motor to drive a right wheel, and

wherein the at least one processor is configured to:

control the left wheel motor and the right wheel motor so that the robot cleaner rotates at the rotation angle and in the rotation direction, and

control the left wheel motor and the right wheel motor so that the robot cleaner travels a forward distance or a backward distance.

11. An operating method of a robot cleaner to avoid a stuck situation through an artificial intelligence (AI), the operating method comprising:

acquiring a surrounding map, based on sensing information acquired through at least one sensor;

determining escape path factors based on the surrounding map by using a compensation model for inferring the escape path factors from a learning surrounding map, when the stuck situation of the robot cleaner is detected; and

controlling a driving motor of the robot cleaner so that the robot cleaner travels, based on the determined escape path factors,

wherein the operating method further comprises:

assigning a reward or a penalty to the compensation model, based on at least one of whether the robot cleaner succeeds in escaping from the stuck situation or an escape time;

assigning a reward of a first value to the compensation model, when the robot cleaner succeeds in escaping from the stuck situation within a preset time; and

assigning a reward of a second value less than the first value to the compensation model when the robot cleaner succeeds in escaping from the stuck situation, after the preset time.

12. The operating method of claim 11 , wherein the escape path factors include at least one of a rotation angle, a rotation direction, a forward direction, or a backward direction.

13. The operating method of claim 12 , wherein the at least one sensor includes:

a bumper sensor configured to sense a bumper event; and

a depth sensor configured to acquire three-dimensional (3D) data, and

wherein the surrounding map is generated by the bumper event and the 3D data.

14. The operating method of claim 13 , wherein the compensation model is a model based on an artificial neural network which is trained through reinforcement learning in a deep learning algorithm or a machine learning algorithm, and

wherein the reinforcement learning is performed through a Markov decision process.

15. The operating method of claim 14 , further comprising:

acquiring the escape path factors based on an output result of the compensation model from the learning surrounding model; and

acquiring, based on the acquired escape path factors, whether the robot cleaner succeeds in escaping from the stuck situation and the escape time taken until the robot cleaner escapes the stuck situation.

16. The operating method of claim 13 , wherein the surrounding map includes at least one obstacle.

17. The operating method of claim 12 , wherein the bumper sensor measures an impulse applied to a bumper provided in the robot cleaner, and generates the bumper event when the impulse is equal to or greater than a preset impulse, and

wherein the operating method further comprises determining the robot cleaner as being in the stuck situation, if the when number of times that the bumper event occurs is equal to or greater than preset number of times.

18. The operating method of claim 12 , wherein the driving motor includes:

a left wheel motor to drive a left wheel; and

a right wheel motor to drive a right wheel,

wherein the controlling of the driving motor includes:

controlling the left wheel motor and the right wheel motor so that the robot cleaner rotates at the rotation angle and in the rotation direction; and

controlling the left wheel motor and the right wheel motor so that the robot cleaner travels a forward distance or a backward distance.

19. A robot cleaner configured to avoid a stuck situation through an artificial intelligence (AI), the robot cleaner comprising:

at least one sensor configured to acquire sensing information;

a driving motor configured to drive the robot cleaner;

a memory configured to store a compensation model to infer escape path factors from a learning surrounding map; and

at least one processor configured to:

acquire a surrounding map, based on the sensing information,

determine the escape path factors based on the surrounding map by using the compensation model, when the stuck situation of the robot cleaner is detected, and

control the driving motor so that the robot cleaner travels, based on the determined escape path factors,

wherein the escape path factors include at least one of a rotation angle, a rotation direction, a forward direction, or a backward direction,

wherein the at least one sensor includes:

a bumper sensor configured to sense a bumper event, and

a depth sensor configured to acquire three-dimensional (3D) data,

wherein the surrounding map is generated by the bumper event and the 3D data,

wherein the bumper sensor measures an impulse applied to a bumper provided in the robot cleaner, and generates the bumper event when the impulse is equal to or greater than a preset impulse, and

wherein the at least one processor is configured to determine the robot cleaner as being in the stuck situation, when number of times that the bumper event occurs is equal to or greater than preset number of times.