Reduced degree of freedom robotic controller apparatus and methods
Apparatus and methods for training and controlling of, for instance, robotic devices. In one implementation, a robot may be trained by a user using supervised learning. The user may be unable to control all degrees of freedom of the robot simultaneously. The user may interface to the robot via a control apparatus configured to select and operate a subset of the robot's complement of actuators. The robot may comprise an adaptive controller comprising a neuron network. The adaptive controller may be configured to generate actuator control commands based on the user input and output of the learning process. Training of the adaptive controller may comprise partial set training. The user may train the adaptive controller to operate first actuator subset. Subsequent to learning to operate the first subset, the adaptive controller may be trained to operate another subset of degrees of freedom based on user input via the control apparatus.
1. A method of training a controller of a robot, the method comprising:
receiving at least one input by the controller, the controller including a plurality of layers forming a multi-layer neuron network, the plurality of layers including an input layer and an output layer, the input layer including a first set of plurality of neurons and the output layer including a second set of plurality of neurons, the first set of plurality of neurons being connected to the second set of plurality of neurons via a plurality of connections;
generating at least one output by the second set of plurality of neurons based on the at least one input; and
transmitting the at least one output to a plurality of actuators, the at least one output configured to effect multiple degrees of freedom of the robot,
wherein each of the plurality of connections of the first set of plurality of neurons and the second set of plurality of neurons being assigned an efficacy value, the efficacy value of the first set of plurality of neurons is modified independently of the efficacy value of the second set of plurality of neurons such that the multiple degrees of freedom of the robot are not trained contemporaneously.
2. A robotic apparatus, comprising:
at least one controller configured to execute computer readable instructions to,
receive at least one input by the controller, the controller including a plurality of layers forming a multi-layer neuron network, the plurality of layers including an input layer and an output layer, the input layer including a first set of plurality of neurons and the output layer including a second set of plurality of neurons, the first set of plurality of neurons being connected to the second set of plurality of neurons via a plurality of connections;
generate at least one output by the second set of plurality of neurons based on the at least one input; and
transmit the at least one output to a plurality of actuators, the at least one output configured to effect multiple degrees of freedom of the robot,
wherein each of the plurality of connections of the first set of plurality of neurons and the second set of plurality of neurons being assigned an efficacy value, the efficacy value of the first set of plurality of neurons is modified independently of the efficacy value of the second set of plurality of neurons such that the multiple degrees of freedom of the robot are not trained contemporaneously.
3. A non-transitory computer readable medium having computer readable instructions stored thereon, that when executed by cause the controller to;
receive at least one input by the controller, the controller including a plurality of layers forming a multi-layer neuron network, the plurality of layers including an input layer and an output layer, the input layer including a first set of plurality of neurons and the output layer including a second set of plurality of neurons, the first set of plurality of neurons being connected to the second set of plurality of neurons via a plurality of connections;
generate at least one output by the second set of plurality of neurons based on the at least one input; and
transmit the at least one output to a plurality of actuators, the at least one output configured to effect multiple degrees of freedom of the robot,
wherein each of the plurality of connections of the first set of plurality of neurons and the second set of plurality of neurons being assigned an efficacy value, the efficacy value of the first set of plurality of neurons is modified independently of the efficacy value of the second set of plurality of neurons such that the multiple degrees of freedom of the robot are not trained contemporaneously.
4. The method of claim 1 , wherein the output layer precedes the input layer such that the second set of plurality of neurons provides a feedback signal to the first set of plurality of neurons.
5. The method of claim 1 , wherein the input layer corresponds to a feature extraction layer that is configured to respond to the at least one input.
6. The method of claim 1 , wherein the efficacy value includes a latency parameter, the latency parameter corresponding to time difference of the at least one input from the first set of plurality of neurons to the second set of plurality of neurons.
7. The method of claim 1 , further comprising:
optimizing performance of the robot by minimizing average value of a performance function; and
adjusting the plurality of connections between the first and second sets of plurality of neurons based on the at least one input.
8. The method of claim 1 , further comprising:
training a first subset of the plurality of actuators to operate in accordance with a first degree of freedom based on the efficacy value; and
training a second subset of plurality of actuators to operate in accordance with a second degree of freedom based on the efficacy value.
9. The method of claim 8 , wherein,
the training of the first subset of plurality of actuators is done at a first time interval and the training of the second subset of plurality of actuators is done at a different second time interval, the second time interval occurs after the first time interval, and
the efficacy value of each of the plurality of connections is updated during the second time interval.
10. The method of claim 1 , further comprising: updating the efficacy value of each of the plurality of connections.
11. The robotic apparatus of claim 2 , wherein the output layer precedes the input layer such that the second set of plurality of neurons provides a feedback signal to the first set of plurality of neurons.
12. The robotic apparatus of claim 2 , wherein the input layer corresponds to a feature extraction layer that is configured to respond to the at least one input.
13. The robotic apparatus of claim 2 , wherein the efficacy value includes a latency parameter, the latency parameter corresponding to time difference of the at least one input from the first set of plurality of neurons to the second set of plurality of neurons.
14. The robotic apparatus of claim 2 , wherein the computer readable instructions further cause the controller to,
optimize performance of the robot by minimizing average value of a performance function; and
adjust the plurality of connections between the first and second sets of plurality of neurons based on the at least one input.
15. The robotic apparatus of claim 2 , wherein the computer readable instructions further cause the controller to,
train a first subset of the plurality of actuators to operate in accordance with a first degree of freedom based on the efficacy value; and
train a second subset of plurality of actuators to operate in accordance with a second degree of freedom based on the efficacy value.
16. The robotic apparatus of claim 15 , wherein,
the training of the first subset of plurality of actuators is done at a first time interval and the training of the second subset of plurality of actuators is done at a different second time interval, the second time interval occurs after the first time interval, and
the efficacy value of each of the plurality of connections is updated during the second time interval.
17. The non-transitory computer readable medium of claim 3 , wherein the output layer precedes the input layer such that the second set of plurality of neurons provides a feedback signal to the first set of plurality of neurons.
18. The non-transitory computer readable medium of claim 3 , wherein the input layer corresponds to a feature extraction layer that is configured to respond to the at least one input.
19. The non-transitory computer readable medium of claim 3 , wherein the efficacy value includes a latency parameter, the latency parameter corresponding to time difference of the at least one input from the first set of plurality of neurons to the second set of plurality of neurons.
20. The non-transitory computer readable medium of claim 3 , wherein the computer readable instructions further cause the controller to,
optimize performance of the robot by minimizing average value of a performance function; and
adjust the plurality of connections between the first and second sets of plurality of neurons based on the at least one input.