Policy layers for machine control
Apparatuses, systems, and techniques provide a policy that can be executed to cause a machine to move. In at least one embodiment, a first policy layer is provided to cause the machine to execute a first motion that causes the machine to accelerate to reach an unbiased state. A second policy layer is provided to cause the machine to execute a second motion without influencing the unbiased state to be reached by machine. The policy can comprise the first and second policy layers.
1. A computer system comprising one or more processors and computer-readable memory storing instructions executable by the one or more processors to cause the computer system to at least:
identify a policy to cause a machine to execute at least one movement, the policy including at least a plurality of policy layers comprising:
a first policy layer to cause the machine to execute a first motion that reaches an unbiased state, the first motion to be limited by at least a first parameter associated with the machine and a second parameter associated with an area in which the machine is to operate, and
a second policy layer to cause the machine to execute a second motion without influencing the unbiased state associated with the first policy layer; and
execute the policy to cause the machine to execute the at least one movement, the at least one movement comprising at least the first motion and the second motion.
2. The computer system according to claim 1 , wherein the first policy layer comprises at least a first nonlinear second-order differential equation and the second policy layer comprises at least a second nonlinear second-order differential equation, the first nonlinear second-order differential equation unbiased to cause the first motion to come to rest and the second nonlinear second-order differential equation unbiased to cause the second motion to come to rest.
3. The computer system according to claim 1 , wherein the first policy layer is a first geometric fabric comprising a nonlinear second-order differential equation and the second policy layer is a second geometric fabric comprising another second-order differential equation.
4. The computer system according to claim 1 , wherein the first parameter comprises at least first data comprising a one or more limits associated with a joint of the machine and the second parameter comprises at least second data comprising a target position to be reached by the machine, the target position located in an Euclidean space associated with the area in which the machine is to operate.
5. The computer system according to claim 1 , wherein the first policy layer is energized by a first Finsler energy and the second policy layer is energized by a second Finsler energy, the first Finsler energy is homogeneous of degree two and the second Finsler energy is homogenous of degree two.
6. The computer system according to claim 1 , wherein the machine is an articulated robot comprising at least one arm, the first policy layer to cause the at least one arm to execute the first motion comprising a straight line movement limited by the first parameter and the second parameter, and wherein the first parameter corresponds to a joint of the at least one arm and the second parameter corresponds to coordinate in the area in which the articulated robot is to operate.
7. The computer system according to claim 1 , wherein the second motion is to occur subsequent to the first motion, the second motion to cause a gripper of the machine to execute a movement based on at least one task that the machine is to undertake.
8. The computer system according to claim 1 , wherein the plurality of policy layers comprises a third policy layer, the third policy layer to cause the machine to execute a third motion without influencing the unbiased state associated with the first policy layer and an unbiased state associated with the second policy layer, and wherein the third motion is to cause the machine to avoid at least one obstacle in the area in which the machine is to operate.
9. A device comprising:
one or more processors and memory storing executable instructions that, as a result of being executed by the one or more processors, cause the device to:
generate a first policy layer to cause a machine to execute a first motion that causes the machine to accelerate to reach an unbiased state, and
generate a second policy layer to cause the machine to execute a second motion without influencing the unbiased state to be reached by machine; and
execute the first policy layer and the second policy layer to cause the machine to execute the first motion and the second motion.
10. The device according to claim 9 , wherein the one or more processors and the memory storing executable instructions that, as a result of being executed by the one or more processors, further cause the device to generate a policy comprising the first policy layer and the second policy layer, the policy to cause the machine to execute at least one action.
11. The device according to claim 10 , wherein the first policy layer and the second policy layer are generated consecutively starting with the first policy layer and followed by the second policy layer.
12. The device according to claim 9 , wherein the first policy layer comprises at least a first nonlinear second-order differential equation and the second policy layer comprises at least a second nonlinear second-order differential equation, the first nonlinear second-order differential equation unbiased to cause the first motion to come to rest and the second nonlinear second-order differential equation unbiased to cause the second motion to come to rest.
13. The device according to claim 9 , wherein the first policy layer is a first geometric fabric comprising a nonlinear second-order differential equation and the second policy layer is a second geometric fabric comprising another second-order differential equation.
14. The device according to claim 9 , wherein the first policy layer comprises at least first data comprising a one or more limits associated with a joint of the machine and the second policy layer comprises at least second data comprising a target position to be reached by the machine, the target position located in an Euclidean space associated with the area in which a portion of the machine is to operate.
15. The device according to claim 9 , wherein the first policy layer is energized by a first Finsler energy and the second policy layer is energized by a second Finsler energy.
16. The device according to claim 9 , wherein the machine is an articulated robot comprising at least one arm, the first policy layer to cause the at least one arm to execute the first motion comprising a substantially straight line movement.
17. The device according to claim 9 , wherein the second motion is to occur subsequent to the first motion, the second motion to cause a gripper of the machine to execute a movement based on at least one task that the machine is to undertake.
18. The device according to claim 9 , wherein the one or more processors and the memory storing executable instructions that, as a result of being executed by the one or more processors, further cause the device to generate a third policy layer, the third policy layer to cause the machine to execute a third motion without influencing the unbiased state associated with the first policy layer and an unbiased state associated with the second policy layer, and wherein the third motion is to cause the machine to avoid at least one obstacle in the area in which the machine is to operate.
19. A computer-implemented method, comprising:
generating a first policy layer to cause a machine to execute a first motion that causes the machine to accelerate to reach an unbiased state, and
generating a second policy layer to cause the machine to execute a second motion without influencing the unbiased state to be reached by machine; and
executing the first policy layer and the second policy layer to cause the machine to execute the first motion and the second motion.
20. The computer-implemented method according to claim 19 , further comprising a third policy layer, the third policy layer to cause the machine to execute a third motion without influencing the unbiased state associated with the first policy layer and an unbiased state associated with the second policy layer, and wherein the third motion is to cause the machine to avoid at least one obstacle in the area in which the machine is to operate.
21. The computer-implemented method according to claim 19 , wherein the first policy layer and the second policy layer are generated consecutively starting with the first policy layer and followed by the second policy layer.
22. The computer-implemented method according to claim 19 , wherein the first policy layer comprises at least a first nonlinear second-order differential equation and the second policy layer comprises at least a second nonlinear second-order differential equation, the first nonlinear second-order differential equation unbiased to cause the first motion to come to rest and the second nonlinear second-order differential equation unbiased to cause the second motion to come to rest.
23. The computer-implemented method according to claim 19 , wherein the first policy layer is a first geometric fabric comprising a nonlinear second-order differential equation and the second policy layer is a second geometric fabric comprising another second-order differential equation.
24. The computer-implemented method according to claim 19 , wherein the first policy layer comprises at least first data comprising a one or more limits associated with a joint of the machine and the second policy layer comprises at least second data comprising a target position to be reached by the machine, the target position located in an Euclidean space associated with the area in which a portion of the machine is to operate.