Object hand-over between robot and actor

1. A robotic device comprising:

an arm comprising a palm and two or more fingers;

a first sensor;

an image sensor disposed within the palm; and

a control system configured to:

receive, from the first sensor, first data associated with an actor;

based on the first data, detect an indication to hand an object over from the robotic device to the actor, wherein the robotic device is holding the object between the two or more fingers; and

in response to detecting the indication to hand the object over from the robotic device to the actor, cause the robotic device to enter a give-object mode, wherein, when the robotic device is in the give-object mode, the control system is further configured to:

receive, from the image sensor, image data representing an environment around the palm, wherein the image data comprises (i) a first area containing the object held between the two or more fingers and (ii) a second area proximate to the object;

when the second area proximate to the object is vacant, cause the arm to move towards the actor; and

when a hand of the actor is detected in the second area, cause (i) the arm to stop and (ii) the two or more fingers to open to release the object into the hand of the actor.

2. The robotic device of claim 1 , wherein, when the robotic device is in the give-object mode, the control system is further configured to:

cause the two or more fingers to open to release the object into the hand of the actor in response to the hand of the actor being detected in the second area for at least a threshold period of time.

3. The robotic device of claim 1 , wherein, when the robotic device is in the give-object mode, the control system is further configured to:

determine, based on the image data, that the second area exceeds a size threshold such that the object obstructs no more than a threshold portion of a field of view of the image sensor; and

cause the arm to move towards the actor further based on determining that the second area exceeds the size threshold.

4. The robotic device of claim 1 , wherein the second area proximate to the object comprises at least one of, (i) an area below the object, (ii) an area above the object, or (iii) an area to a side of the object.

5. The robotic device of claim 1 , wherein, when the robotic device is in the give-object mode, the control system is further configured to:

detect the hand in the second area by comparing an intensity of at least a portion of pixels in the second area to a threshold intensity.

6. The robotic device of claim 1 , wherein, when the robotic device is in the give-object mode, the control system is further configured to:

determine a baseline intensity of pixels in the second area; and

detect the hand in the second area by comparing an intensity of at least a portion of pixels in the second area to the baseline intensity.

7. The robotic device of claim 1 , wherein the image data comprises a third area proximate to the object that is expected to be vacant, and wherein, when the robotic device is in the give-object mode, the control system is further configured to:

detect the hand in the second area by comparing an intensity of at least a portion of pixels in the second area to an intensity of at least a portion of pixels in the third area.

8. The robotic device of claim 1 , wherein the first sensor comprises an additional image sensor, and wherein, when the robotic device is in the give-object mode, the control system is further configured to:

determine that the second area is smaller than a threshold size due to the object occluding a field of view of the image sensor; and

in response to determining that the second area is smaller than the threshold size and based on additional image data received from the additional image sensor, (i) cause the arm to move towards the actor, (ii) cause the arm to stop, and (iii) cause the two or more fingers to open to release the object into the hand of the actor, wherein the additional image data represents the environment around the palm from a first perspective different from a second perspective of the image sensor disposed within the palm.

9. The robotic device of claim 1 , further comprising a depth sensor disposed within the palm, wherein, when the robotic device is in the give-object mode, the control system is further configured to:

receive, from the depth sensor, depth data representing (i) a first distance between the palm and physical features within the first area and (ii) a second distance between the palm and physical features in the second area;

when (i) the second area proximate to the object is vacant and (ii) the second distance is greater than the first distance, cause the arm to move towards the actor; and

when the hand of the actor is detected in the second area and the second distance is smaller than the first distance, cause the arm to stop and the two or more fingers to open to release the object into the hand of the actor.

10. The robotic device of claim 1 , wherein, when the robotic device is in the give-object mode, the control system is further configured to:

move the arm along a predetermined trajectory to gesture that the robotic device is in the give-object mode and is ready to hand the object over to the actor.

11. The robotic device of claim 1 , further comprising an additional image sensor, wherein, when the robotic device is in the give-object mode, the control system is further configured to:

receive, from the additional image sensor, additional image data representing a relative position between the arm and the hand;

based on the additional image data, determine a trajectory for moving the arm towards the hand; and

when the second area proximate to the object is vacant, causing the arm to move towards the actor along the trajectory.

12. The robotic device of claim 1 , wherein the first sensor comprises an additional image sensor, wherein the first data comprises additional image data, and wherein the control system is configured to detect the indication to hand the object over from the robotic device to the actor by:

identifying, based on the additional image data, a virtual skeleton representing the actor; and

detecting the indication to hand the object over from the robotic device to the actor based on a pose of the virtual skeleton reaching a predetermined configuration.

13. The robotic device of claim 1 , wherein the first sensor comprises a microphone communicatively connected to the control system, and wherein the control system is configured to detect the indication to hand the object over from the robotic device to the actor by:

detecting, by way of the microphone, an utterance by the actor; and

determining that the utterance indicates to hand the object over from robotic device to the actor.

14. The robotic device of claim 1 , wherein the first sensor comprises an additional image sensor, wherein the first data comprises additional image data, wherein the robotic device further comprises an audio speaker and a microphone, and wherein the control system is configured to detect the indication to hand the object over from the robotic device to the actor by:

identifying, based on the additional image data, a virtual skeleton representing the actor;

in response to a pose of the virtual skeleton reaching a predetermined configuration, causing the audio speaker to generate an utterance requesting reception of the object from the arm by the actor; and

detecting, by way of the microphone, an utterance by the actor indicating to hand the object over from the robotic device to the actor.

15. The robotic device of claim 1 , wherein the first sensor comprises an additional image sensor, and wherein, when the robotic device is in the give-object mode, the control system is further configured to:

receive, from the additional image sensor, additional image data representing the actor reaching for the object; and

based on the additional image data, verifying that the image data received from the image sensor disposed within the palm represents the hand of the actor in the second area.

16. A method comprising:

receiving, from a first sensor on a robotic device and by a control system of the robotic device, first data associated with an actor;

based on the first data, detecting, by the control system, an indication to hand an object over from the robotic device to the actor, wherein the robotic device is holding the object between two or more fingers of an arm; and

in response to detecting the indication to hand the object over from the robotic device to the actor, operating the robotic device in a give-object mode, wherein operating in the give-object mode comprises:

receiving, from an image sensor disposed within a palm of the arm and by the control system, image data representing an environment around the palm, wherein the image data comprises (i) a first area containing the object held between the two or more fingers and (ii) a second area proximate to the object;

when the second area proximate to the object is vacant, causing, by the control system, the arm to move towards the actor; and

when a hand of the actor is detected in the second area, causing, by the control system, (i) the arm to stop and (ii) the two or more fingers to open to release the object into the hand of the actor.

17. The method of claim 16 , wherein operating in the give-object mode further comprises:

causing the two or more fingers to open to release the object into the hand of the actor in response to the hand of the actor being detected in the second area for at least a threshold period of time.

18. The method of claim 16 , wherein the image data comprises a third area proximate to the object that is expected to be vacant, and wherein operating in the give-object mode further comprises:

detecting the hand in the second area by comparing an intensity of at least a portion of pixels in the second area to an intensity of at least a portion of pixels in the third area.

19. The method of claim 16 , wherein the robotic device further comprises a depth sensor disposed within the palm, and wherein operating in the give-object mode further comprises:

receiving, from the depth sensor, depth data representing (i) a first distance between the palm and physical features within the first area and (ii) a second distance between the palm and physical features in the second area;

when (i) the second area proximate to the object is vacant and (ii) the second distance is greater than the first distance, causing the arm to move towards the actor; and

when the hand of the actor is detected in the second area and the second distance is smaller than the first distance, causing the arm to stop and the two or more fingers to open to release the object into the hand of the actor.

20. A non-transitory computer readable medium having stored thereon instructions that, when executed by a computing device, cause the computing device to perform operations comprising:

receiving, from a first sensor on a robotic device, first data associated with an actor;

based on the first data, detecting an indication to hand an object over from the robotic device to the actor, wherein the robotic device is holding the object between two or more fingers of an arm; and

in response to detecting the indication to hand the object over from the robotic device to the actor, operating the robotic device in a give-object mode, wherein operating in the give-object mode comprises:

receiving, from an image sensor disposed within a palm of the arm, image data representing an environment around the palm, wherein the image data comprises (i) a first area containing the object held between the two or more fingers and (ii) a second area proximate to the object;

when the second area proximate to the object is vacant, causing the arm to move towards the actor; and

when a hand of the actor is detected in the second area, causing (i) the arm to stop and (ii) the two or more fingers to open to release the object into the hand of the actor.