APPARATUS AND METHODS FOR OPERATING ROBOTIC DEVICES USING SELECTIVE STATE SPACE TRAINING
Apparatus and methods for training and controlling of e.g., robotic devices. In one implementation, a robot may be utilized to perform a target task characterized by a target trajectory. The robot may be trained by a user using supervised learning. The user may interface to the robot, such as via a control apparatus configured to provide a teaching signal to the robot. The robot may comprise an adaptive controller comprising a neuron network, which may be configured to generate actuator control commands based on the user input and output of the learning process. During one or more learning trials, the controller may be trained to navigate a portion of the target trajectory. Individual trajectory portions may be trained during separate training trials. Some portions may be associated with robot executing complex actions and may require additional training trials and/or more dense training input compared to simpler trajectory actions.
1 .- 26 . (canceled)
27 . A method of operating a robotic controller apparatus, the method comprising:
determining by the robotic controller apparatus a current performance measure of the robotic controller apparatus associated with performing a target task autonomously along a target trajectory;
selecting by the robotic controller apparatus a first portion of the target trajectory, the first portion being characterized by a first performance measure that is lower as compared to a second performance measure associated with another portion of the target trajectory; and
receiving a teaching input for navigating the first portion of the target trajectory, the teaching input being configured to navigate the robot towards the target trajectory and improve the current performance measure.
28 . The method of claim 27 , wherein the robotic controller apparatus is operable in accordance with a supervised learning process configured based at least in part on the teaching input, the supervised learning process being adapted based at least in part on the current performance measure.
29 . The method of claim 27 , wherein the first portion of the target trajectory is characterized by a state space, and an extent of the state space is characterized by a first dimension having a first value and by a second dimension having a second value.
30 . The method of claim 29 , further comprising performing the target task autonomously, wherein:
the performance of the target task by the robotic controller apparatus comprises a provision of a control signal by the robotic controller apparatus to a robotic platform; and
the first dimension is selected from the group consisting of a spatial coordinate, a velocity, an acceleration, and an orientation of the platform.
31 . The method of claim 29 , wherein:
the determining the first portion of the target trajectory is based at least on the first dimension being outside a target range of at least one state space parameter.
32 . The method of claim 27 , wherein the first performance measure and second performance measure are determined based at least in part on a deviation of an actual position of the robotic controller apparatus from the target trajectory.
33 . The method of claim 27 , wherein the robotic controller apparatus is configured to operate in accordance with a supervised learning process based at least on the teaching input and a plurality of training trials, the supervised learning process being adapted based on the current performance measure.
34 . The method of claim 27 , wherein the teaching input comprises a control signal for a controllable degree of freedom of motion of the robotic controller apparatus.
35 . A robot comprising:
a platform configured to navigate an environment;
a sensor module configured to provide information related to the environment of the platform; and
an adaptive controller configured to:
provide navigation instructions to the platform based at least in part on the information provided by the sensor module;
receive one or more teaching inputs providing instructions for the navigation of the platform;
determine a current performance measure of the platform associated with navigating autonomously along a target trajectory;
select a first portion of the target trajectory, the first portion being characterized by a first performance measure that is lower as compared to a second performance measure associated with another portion of the target trajectory; and
receive a first teaching input for navigating the first portion of the target trajectory, the first teaching input being configured to provide instructions to navigate the robot towards the target trajectory and improve the current performance measure.
36 . The robot of claim 35 , further comprising a user interface configured to receive the first teaching input.
37 . The robot of claim 36 , wherein the user interface is remotely located from the robot and communicatively coupled with the robot.
38 . The robot of claim 35 , further comprising one or more actuators configured to actuate the robot along one or more controllable degrees of freedom of motion.
39 . The robot of claim 35 , wherein the adaptive controller is further configured to: request assistance for navigating the first portion of the target trajectory, wherein receiving the first teaching input is in response to the request for assistance.
40 . The robot of claim 35 , wherein the first portion comprises a state space extent that is less than half of a state space extent associated with traversing all of the target trajectory.
41 . An adaptive controller apparatus comprising a plurality of computer readable instructions configured to, when executed, cause a performance of a target task, the computer readable instructions configured to cause the adaptive controller apparatus to:
receive information relating to autonomous navigation of a robotic device;
determine from the information a current performance measure of the robotic device associated with navigating autonomously along a target trajectory;
select a first portion of the target trajectory, the first portion being characterized by a first performance measure that is lower as compared to a second performance measure associated with another portion of the target trajectory; and
receive a first teaching input for navigating the first portion of the target trajectory, the first teaching input being configured to provide instructions to navigate the robotic device towards the target trajectory and improve the current performance measure.
42 . The apparatus of claim 41 , wherein the computer readable instructions are further configured to cause the adaptive controller apparatus to perform the target task autonomously by sending a control signal to the robotic device.
43 . The apparatus of claim 41 , wherein the computer readable instructions are further configured to cause the adaptive controller apparatus to perform a supervised learning process over a plurality of trials, the supervised learning process being adapted based at least in part on the first teaching input and the current performance measure.
44 . The apparatus of claim 41 , wherein the teaching input comprises a control signal for a controllable degree of freedom of motion of the adaptive controller apparatus.
45 . The apparatus of claim 41 , wherein:
the received information provides a context to the adaptive controller apparatus; and
the adaptive controller apparatus determines a navigation control signal associated with the context and sends the control signal to the robotic device.
46 . The apparatus of claim 41 , wherein the target task comprises a cleaning task.