Robotically negotiating stairs

1. A method comprising:

receiving, at data processing hardware, sensor data from one or more sensor(s) of a robot, the sensor data corresponding to an environment with a stair, the stair comprising a tread surface, the robot comprising at least four legs;

determining, by the data processing hardware, a step region of the stair based on the sensor data, the step region identifying a subset of the tread surface of the stair for placement of a distal end of a leg of the at least four legs;

determining, by the data processing hardware, a step location within the step region of the stair; and

instructing movement, by the data processing hardware, of the distal end of the leg to the step location within the step region of the stair.

2. The method of claim 1 , further comprising:

determining, by the data processing hardware, the robot is descending a set of stairs with one or more front sensors of a front portion of the robot oriented in a traversal direction of the robot, the set of stairs comprising the stair; and

shifting a weight distribution of the robot toward the front portion of the robot for traversing the set of stairs.

3. The method of claim 1 , further comprising:

instructing, by the data processing hardware, the robot to descend a set of stairs with one or more front sensors of a front portion of the robot oriented in a direction opposite to a traversal direction of the robot, the set of stairs comprising the stair.

4. The method of claim 1 , wherein the step region excludes a portion of the tread surface of the stair from the subset of the tread surface of the stair for placement of the distal end of the leg, wherein the portion of the tread surface of the stair is within a particular proximity of an edge of the stair.

5. The method of claim 1 , further comprising:

switching, by the data processing hardware, from a first cadence of the robot to a second cadence of the robot based on detecting the stair in the sensor data.

6. A method comprising:

receiving, at data processing hardware, sensor data from one or more sensor(s) of a robot, the sensor data corresponding to an environment with a stair, the robot comprising:

a back portion,

a front portion, and

at least four legs, each of the at least four legs having a knee joint and a distal end, each of the knee joints oriented further toward the back portion than the corresponding distal end;

determining, by the data processing hardware, a step region of the stair based on the sensor data, the step region representing a subset of a tread surface of the stair;

determining, by the data processing hardware, a step location within the step region of the stair; and

instructing, by the data processing hardware, based on the step location within the step region of the stair, the robot to descend the stair such that the back portion precedes the front portion down the stair.

7. The method of claim 6 , wherein the robot is configured to ascend the stair such that front portion precedes the back portion up the stair.

8. The method of claim 6 , wherein the robot is capable of traversing a non-stair portion of the environment forwards.

9. The method of claim 6 , further comprising:

switching, by the data processing hardware, from a first cadence of the robot to a second cadence of the robot based on detecting the stair in the sensor data.

10. The method of claim 6 , wherein the one or more sensor(s) include stereo cameras on the front portion of the robot.

11. The method of claim 6 , wherein instructing the robot to descend the stair comprises:

instructing movement of a distal end of a leg of the at least four legs to the step location within the step region.

12. The method of claim 6 , wherein instructing the robot to descend the stair comprises:

instructing movement of a distal end of a leg of the at least four legs to the step location within the step region in a cadence selected based on identifying a staircase within the environment by reference to the sensor data, the staircase including the stair.

13. The method of claim 6 , wherein the step region identifies the subset of the tread surface of the stair for placement of a distal end of a leg of the at least four legs.

14. The method of claim 6 , further comprising:

adjusting, by the data processing hardware:

a height of the robot with respect to a surface underneath the robot to obtain an adjusted height of the robot; or

a pitch of the robot about an axis of the robot to obtain an adjusted pitch of the robot,

wherein instructing the robot to descend the stair comprises:

instructing the robot to descend the stair according to the adjusted height or the adjusted pitch.

15. The method of claim 6 , further comprising:

detecting, by the data processing hardware, an impact by a leg of the at least four legs; and

shifting, by the data processing hardware, a distal end of the leg to an adjusted step location within the step region based on the impact, wherein the adjusted step location is shifted relative to the step location within the step.

16. The method of claim 6 , further comprising:

detecting, by the data processing hardware, a trip condition of a leg of the at least four legs associated with instability of the robot; and

elevating, by the data processing hardware, a distal end of the leg based on the trip condition.

17. A method comprising:

receiving, at data processing hardware, sensor data from one or more sensor(s) of a robot, the sensor data corresponding to an environment with at least one stair, the robot comprising at least two legs;

detecting, by the data processing hardware, an impact of at least a portion of a leg of the at least two legs with the at least one stair based on the sensor data;

identifying, by the data processing hardware, a response based on the impact; and

instructing, by the data processing hardware, the robot to traverse the at least one stair according to the response.

18. The method of claim 17 , further comprising:

switching, by the data processing hardware, from a first cadence of the robot to a second cadence of the robot based on detecting the at least one stair in the sensor data.

19. The method of claim 17 , wherein the leg comprises a trailing leg during traversal of the at least one stair, the portion of the leg comprises a knee joint, and the response includes moving the knee joint in a traversal direction beneath a body of the robot.

20. The method of claim 17 , wherein instructing the robot to traverse the at least one stair comprises:

instructing movement of the at least two legs according to a cadence selected based on identifying a staircase within the environment by reference to the sensor data, the staircase including the at least one stair.

21. The method of claim 17 , further comprising:

determining, by the data processing hardware, a step region of the at least one stair based on the sensor data, the step region identifying a subset of the at least one stair for placement of a distal end of the leg and excluding a no-step region of the at least one stair.

22. The method of claim 17 , wherein the robot is capable of traversing the environment backwards and forwards.

23. The method of claim 17 , further comprising:

adjusting, by the data processing hardware:

a height of the robot with respect to a surface underneath the robot to obtain an adjusted height of the robot; or

a pitch of the robot about an axis of the robot to obtain an adjusted pitch of the robot,

wherein instructing the robot to traverse the at least one stair comprises:

instructing the robot to traverse the at least one stair according to the adjusted height or the adjusted pitch.

24. A method comprising:

instructing a first step, by data processing hardware, of a distal end of a leg of at least two legs of a robot according to a first step frequency of the robot;

receiving, at the data processing hardware, sensor data from one or more sensor(s) of the robot, the sensor data corresponding to an environment with a stair;

identifying, by the data processing hardware, based on the sensor data, the stair and a subset of a tread surface of the stair;

determining, by the data processing hardware, a step location within the subset of the tread surface of the stair;

switching, by the data processing hardware, from the first step frequency to a second step frequency of the robot based on identifying the stair; and

instructing a second step, by the data processing hardware, of the distal end of the leg to the step location within the subset of the tread surface of the stair according to the second step frequency.

25. The method of claim 24 , further comprising:

instructing, by the data processing hardware the robot to descend the stair such that a back portion of the robot precedes a front portion of the robot down the stair.

26. The method of claim 24 , wherein each of the at least two legs have a knee joint and a distal end, each of the knee joints oriented further toward a back portion of the robot than the corresponding distal end, the method further comprising:

instructing, by the data processing hardware the robot to descend the stair such that the back portion precedes a front portion of the robot down the stair.

27. The method of claim 24 , further comprising:

selecting, by the data processing hardware, a movement controller among multiple movement controllers for traversing the stair, wherein the selected movement controller is associated with the second step frequency.

28. The method of claim 24 , further comprising:

determining, by the data processing hardware, a step region based on the sensor data, the step region identifying a subset of the stair for placement of the distal end of the leg, the step region excluding a no-step region of the stair.

29. The method of claim 24 , wherein the robot is capable of traversing a non-stair portion of the environment forwards.

30. The method of claim 24 , further comprising:

instructing, by the data processing hardware, the robot to descend a set of stairs with one or more front sensors of a front portion of the robot oriented in a traversal direction of the robot, the set of stairs comprising the stair; and

shifting, by the data processing hardware, a weight distribution of the robot toward the front portion of the robot for traversing the set of stairs.