Footstep contact detection
A method of footstep contact detection includes receiving joint dynamics data for a swing phase of a swing leg of the robot, receiving odometry data indicative of a pose of the robot, determining whether an impact on the swing leg is indicative of a touchdown of the swing leg based on the joint dynamics data and an amount of completion of the swing phase, and determining when the impact on the swing leg is not indicative of the touchdown of the swing leg, a cause of the impact based on the joint dynamics data and the odometry data.
1. A method of footstep contact detection, the method comprising:
receiving, at data processing hardware of a robot, joint dynamics data associated with a leg of the robot;
receiving, at data processing hardware of the robot, odometry data indicative of a pose of the robot;
determining, by the data processing hardware, based on the joint dynamics data and an amount of completion of a swing phase of the leg, a classification of an impact on the leg from a set of classifications, wherein the set of classifications comprises a touchdown classification and a non-touchdown classification;
determining, by the data processing hardware, using the joint dynamics data and the odometry data and based on the classification of the impact corresponding to the non-touchdown classification, that occurrence of the impact is based on a particular condition, wherein-based on the joint dynamics data and the odometry data are indicative of the particular condition; and
instructing, by the data processing hardware, movement by the robot based on determining that the occurrence of the impact is based on the particular condition.
2. The method of claim 1 , wherein determining that the occurrence of the impact is based on the particular condition comprises:
determining, further based on determining that a value associated with a joint of the leg exceeds a threshold, that the occurrence of the impact is based on the particular condition.
3. The method of claim 1 , wherein a joint of the leg comprises an abduction-adduction component and a flexion-extension component, wherein determining that the occurrence of the impact is based on the particular condition comprises:
determining, further based on comparing a first value associated with the abduction-adduction component to a second value associated with the flexion-extension component, that the occurrence of the impact is based on the particular condition.
4. The method of claim 1 , wherein determining that the occurrence of the impact is based on the particular condition comprises:
determining, further based on comparing a first torque measurement to one or more predicted torques, that the occurrence of the impact is based on the particular condition.
5. The method of claim 1 , wherein determining that the occurrence of the impact is based on the particular condition comprises:
determining, further based on determining that the leg crossed a second leg of the robot, that the occurrence of the impact is based on the particular condition.
6. The method of claim 5 , further comprising:
instructing movement by the robot to uncross the leg and the second leg.
7. The method of claim 1 , wherein the swing phase comprises a lift-off from a ground surface, a flexion at a knee joint of the leg, an extension of the knee joint, and a touchdown of the leg to the ground surface.
8. The method of claim 1 , wherein determining that the occurrence of the impact is based on the particular condition comprises:
determining, further based on detecting a liftoff scuffing of the leg, that the occurrence of the impact is based on the particular condition,
the method further comprising:
instructing movement by the leg to complete the swing phase.
9. The method of claim 1 , wherein the joint dynamics data indicates one or more of an angle associated with a joint of the leg, a speed associated with the joint, or a torque associated with the joint.
10. The method of claim 1 , wherein determining that the occurrence of the impact is based on the particular condition comprises:
determining, further based on determining a distance between the leg and a second leg of the robot, that the occurrence of the impact is based on the particular condition.
11. The method of claim 1 , wherein determining that the occurrence of the impact is based on the particular condition comprises:
determining, further based on detecting a torque, that the occurrence of the impact is based on the particular condition.
12. A robot comprising:
two or more legs; and
a control system in communication with the two or more legs, the control system comprising data processing hardware configured to:
receive joint dynamics data associated with a leg of the two or more legs;
receive odometry data indicative of a pose of the robot;
determine, based on the joint dynamics data and an amount of completion of a swing phase of the leg, a classification of an impact on the leg from a set of classifications, wherein the set of classifications comprises a touchdown classification and a non-touchdown classification;
determine, using the joint dynamics data and the odometry data and based on the classification of the impact corresponding to the non-touchdown classification, that occurrence of the impact is based on a particular condition, wherein the joint dynamics data and the odometry data are indicative of the particular condition; and
instruct movement by the robot based on determining that the occurrence of the impact is based on the particular condition.
13. The robot of claim 12 , wherein to determine that the occurrence of the impact is based on the particular condition, the data processing hardware is further configured to:
determine, further based on determining that a value associated with a joint of the leg exceeds a threshold, that the occurrence of the impact is based on the particular condition.
14. The robot of claim 12 , wherein a joint of the leg comprises an abduction-adduction component and a flexion-extension component, wherein to determine that the occurrence of the impact is based on the particular condition, the data processing hardware is further configured to:
determine, further based on comparing a first value associated with the abduction-adduction component to a second value associated with the flexion-extension component, that the occurrence of the impact is based on the particular condition.
15. The robot of claim 12 , to determine that the occurrence of the impact is based on the particular condition, the data processing hardware is further configured to:
determine, further based on comparing a first torque measurement to one or more predicted torques, that the occurrence of the impact is based on the particular condition.
16. The robot of claim 12 , wherein to determine that the occurrence of the impact is based on the particular condition, the data processing hardware is further configured to:
determine, further based on determining that the leg crossed a second leg of the two or more legs, that the occurrence of the impact is based on the particular condition,
wherein the data processing hardware is further configured to:
instruct movement by the robot to uncross the leg and the second leg.
17. The robot of claim 12 , wherein the swing phase comprises a lift-off from a ground surface, a flexion at a knee joint of the leg, an extension of the knee joint, and a touchdown of the leg to the ground surface.
18. The robot of claim 12 , wherein to determine that the occurrence of the impact is based on the particular condition, the data processing hardware is further configured to:
determine, further based on detecting a liftoff scuffing of the leg, that the occurrence of the impact is based on the particular condition,
wherein the data processing hardware is further configured to;
instruct movement by the leg to complete the swing phase.
19. The robot of claim 12 , wherein to determine that the occurrence of the impact is based on the particular condition, the data processing hardware is further configured to:
determine, further based on determining a distance between the leg and a second leg of the two or more legs, that the occurrence of the impact is based on the particular condition.
20. The robot of claim 12 , wherein to determine that the occurrence of the impact is based on the particular condition, the data processing hardware is further configured to:
determine, further based on detecting a torque, that the occurrence of the impact is based on the particular condition.
21. A method of footstep contact detection, the method comprising:
receiving, at data processing hardware of a robot, joint dynamics data associated with a leg of the robot;
receiving, at data processing hardware of the robot, odometry data indicative of a pose of the robot;
determining, by the data processing hardware, based on the joint dynamics data and an amount of completion of a swing phase of the leg, that a classification of a first impact on the leg corresponds to a touchdown classification;
determining, by the data processing hardware, that a classification of a second impact on the leg corresponds to a non-touchdown classification, wherein occurrence of the second impact is based on a particular condition, and wherein the joint dynamics data and the odometry data are indicative of the particular condition; and
instructing, by the data processing hardware, movement by the robot based on one or more of determining the classification of the first impact or determining the classification of the second impact.