LOCALIZATION BY LEARNING OF WAVE-SIGNAL DISTRIBUTIONS

• US 20110125323A1
• Filed: 11/05/2010
• Published: 05/26/2011
• Est. Priority Date: 11/06/2009
• Status: Active Grant

• Current Assignee: iRobot Corporation
• Original Assignee: Evolution Robotics Inc. (iRobot Corporation)
• Inventors: Gutmann, Steffen, Fong, Philip, Eade, Ethan, Munich, Mario
• Granted Patent: US 8,930,023 B2
• Time in Patent Office: Days
• US Class Current: 700/258
• CPC Class Codes: B25J 9/1664 characterised by motion, path, trajectory planning
• CPC Class Codes: B25J 9/1694 characterised by use of sensors other than normal servo-feedback from position, speed or acceleration sensors, perception control, multi-sensor controlled systems, sensor fusion
• CPC Class Codes: G05D 1/0234 using optical markers or beacons optical beacons per se G01S1/70
• CPC Class Codes: G05D 1/0261 using magnetic plots
• CPC Class Codes: G05D 1/027 comprising intertial navigation means, e.g. azimuth detector inertial navigation G01C21/16; inertial navigation combined with non-inertial navigation instruments G01C21/165
• CPC Class Codes: G05D 1/0272 comprising means for registering the travel distance, e.g. revolutions of wheels measuring distance traversed on the ground by vehicles, e.g. using odometers G01C22/00
• CPC Class Codes: G05D 1/0274 using mapping information stored in a memory device navigation using map-matching G01C21/30
• CPC Class Codes: G05D 1/028 using a RF signal
• CPC Class Codes: Y10S 901/01 Mobile robot
• CPC Class Codes: Y10S 901/09 Closed loop, sensor feedback controls arm movement
• CPC Class Codes: Y10S 901/46 Sensing device

Abstract

A robot having a signal sensor configured to measure a signal, a motion sensor configured to measure a relative change in pose, a local correlation component configured to correlate the signal with the position and/or orientation of the robot in a local region including the robot'"'"'s current position, and a localization component configured to apply a filter to estimate the position and optionally the orientation of the robot based at least on a location reported by the motion sensor, a signal detected by the signal sensor, and the signal predicted by the local correlation component. The local correlation component and/or the localization component may take into account rotational variability of the signal sensor and other parameters related to time and pose dependent variability in how the signal and motion sensor perform. Each estimated pose may be used to formulate new or updated navigational or operational instructions for the robot.

Claims

1. A system for determining a mobile object'"'"'s pose on a surface, the system comprising:

an object body comprising a signal sensor and a motion sensor;

a drive mechanism configured to move the object body and change the orientation of the object body, so that the object body can have an earlier pose comprising an earlier location and an earlier orientation at an earlier time and a later pose comprising a later location and a later orientation at a later time,

a processing unit configured to logically define the surface as a first plurality of logical cells, wherein a given cell in the first plurality of logical cells is defined, at least in part, by a set of three or more nodes having respective pre-defined locations;

the signal sensor configured to detect and measure a property of one or more light signals from at least a first signal source and to output data corresponding to a value measured, wherein the signal sensor has a substantially fixed orientation relative to the object body, and the signal sensor has a rotational variability such that a first measure of the property of a light signal at a time, measured when the object body is at a location and has a first orientation is different from a second measure of the property of the light signal at the same time measured when the object body is at the same location and has a second orientation different from the first orientation;

the motion sensor configured to measure one or more dead-reckoning properties and to output data corresponding to one or more values of the measured one or more dead-reckoning properties;

a reckoning data processing unit in communication with the motion sensor and configured to calculate a change in orientation, a relative direction of travel, and a relative distance of travel between a first time and a second time based at least in part on a portion of the data from the motion sensor output at the second time;

a calibration data processing unit configured to calculate a first parameter approximating the rotational variability of the signal sensor and a second parameter relating the property of the signal to a respective location, such that the signal property has an expected measure at the respective location and the expected measure can be determined from the second parameter and the respective location, the first parameter and second parameter calculated based at least in part on an initial pose of the object body at an initial time, at least a portion of the data from the signal sensor at the initial time, data calculated by the reckoning data processing unit at the initial time, at least a portion of the data from the signal sensor at a subsequent time, and data calculated by the reckoning data processing unit at a subsequent time;

a cell initialization data processing unit configured to initialize one or more nodes in a set of nodes defining a corresponding cell by calculating an expected measure of the signal property at the respective locations of the one or more nodes, the expected measure calculated based at least in part on the second parameter and the predefined locations of the respective nodes;

a localization and mapping data processing unit configured to use a SLAM algorithm to calculate a refined pose approximating a third pose of the mobile object at a third time within a cell having initialized nodes.

2. The system of claim 1, wherein the calibration data processing unit calculates the first and second parameters using a RANSAC (Random Sample Consensus) algorithm.

3. The system of claim 1, wherein the localization and mapping data processing unit is configured to calculate the refined pose by applying the SLAM algorithm to at least 1) the first parameter 2) output from the reckoning data processing unit at the third time 3) an expected measure of the property of the signal at the refined pose, the expected measure determined at least in part from the initialized expected measures of the property at two or more nodes of the cell, and 4) a measure of the signal by the signal sensor at the third time.

4. The system of claim 1, wherein the mobile object is less than 6 centimeters in height, less than 0.5 meters in length and less than 0.5 meters in width.

5. The system of claim 1, where the mobile object is an autonomous robot.

6. The system of claim 1, where the first signal source is not a ceiling mounted light fixture.

7. The system of claim 1, where the detected light signal is one of an ultraviolet light signal or an infrared light signal.

8. The system of claim 1, where the detected light signal is reflected from a ceiling.

9. A method for determining a pose (location and orientation) of a mobile device, the method comprising:

dividing a representation of a substantially planar operating environment into a first plurality of cells, wherein a given cell in the first plurality of cells is defined by a set of three or more nodes having respective locations;

associating at least a first node with properties comprising a location and at least one parameter related to a detected signal detected by the mobile device;

computing an estimated value of at least one property of the signal at an estimated pose of the mobile device at an actual pose in a first cell, the computing based at least in part on the estimated location of the estimated pose and one or more of the respective properties associated with two or more of the nodes defining the first cell; and

revising the estimated pose based at least in part on;

the estimated pose, the estimated value of the signal property, and a value of the signal property measured at the actual pose.

10. The method of claim 9 wherein the detected signal is a transmitted radio frequency signal or a light signal.

11. The method of claim 10 wherein the detected signal is a digital signal transmitted by radio.

12. The method of claim 10 wherein the detected signal is a digital signal used in wireless communications.

13. The method of claim 9 wherein the at least one signal property is time invariant relative to a location.

14. The method of claim 9 wherein computing of the estimated value is further based at least in part on the estimated orientation of the estimated pose and a rotational variation parameter.

15. The method of claim 14 further comprising calculating a rotational variation parameter.

16. The method of claim 9 wherein revising the estimated pose further comprises revising at least one of the parameters associated with at least one of the nodes defining the first cell.

17. The method of claim 9 wherein revising the estimated pose comprises applying a SLAM implementation to at least the estimated pose, the estimated value of the signal property, and the value of the signal property measured at the actual pose.

18. The method of claim 9 wherein revising the estimated pose occurs when the value of the signal property at the actual pose is measured.

19. The method of claim 18 wherein the value of the signal property at the actual pose is not retained after a value of the signal property at a second actual pose is measured.

20. The method of claim 9 further comprising preparing instructions to move the mobile device based at least in part on the revised estimated pose.

21. The method of claim 9 wherein associating a node with properties comprises:

associating the node with a node location;

measuring a property of the signal at a first pose of the mobile device, the first pose located within a first cell defined in part by the node;

moving the mobile device to a second pose in the first cell;

measuring the signal property at the second pose;

obtaining a measure of the relative change in location from the first pose to the second pose, said measure not determined by measuring the signal property;

learning a relationship between the signal property and location within the cell;

estimating a value of the signal property at the node location based at least on the node location and the learned relationship between the signal property and location; and

associating the node with the estimated value of the signal property.

22. The method of claim 21 wherein associating a node with properties further comprises:

obtaining a measure of the relative change in orientation from the first pose to the second pose, said measure not determined by measuring the signal property; and

learning a relationship between the signal property and orientation of the mobile device at a location within the cell; and

wherein estimating a value of the signal property at the node location is further based on the learned relationship between the signal property and orientation; and

wherein the node is further associated with a parameter indicative of orientation.

23. The method of claim 9 wherein the location of the first node is determined based at least in part on the location of the mobile device at the time the node'"'"'s location is determined.

24. The method of claim 9 wherein the location of the first node is determined based at least in part on the location of a second node.

25. The method of claim 9 wherein the respective nodes are located at predefined distances from each other.

26. The method of claim 9, further comprising emitting a signal configured to be detected by the mobile device.

27. The method of claim 9, wherein the at least one parameter related to a signal and associated with a node comprises one or more parameters for an equation modeling a property of the signal within a cell defined in part by the node.

28. The method of claim 27 wherein the equation is a linear interpolation.

29. The method of claim 27 wherein the equation is a Spline.

30. The method of claim 28, wherein the Spline is a Nurbs.

31. The method of claim 22, further comprising moving the mobile device from a third pose in the first cell to a fourth pose in the first cell;

and wherein

computing an estimated value of the signal property at an estimate of the fourth pose of the mobile device comprises interpolating based on a measure of the distance of the estimated pose from the respective nodes of the set of nodes defining the first cell and the estimated value of the signal property associated with the respective nodes of the set;

revising the estimated fourth pose comprises applying a SLAM implementation to a value of the signal property as measured at the fourth pose;

the estimated value of the signal property; and

a measure, not determined by measuring the signal property, of an estimated change in location from the third pose to the fourth pose.

32. The method of claim 31, wherein computing the estimated value of the signal property at the estimated pose is further based at least in part on the relationship between the signal property and mobile device orientation.

33. The method of claim 9, wherein the, dividing, associating, computing, and revising steps are performed by elements of the mobile device.

34. The method of claim 31, wherein the dividing, associating, computing, revising, obtaining, learning, associating, and moving steps are performed by elements of the mobile device.

35. The method of claim 27, wherein one or more of the learning or estimating steps are performed by elements distinct from the mobile device.

36. The method of claim 9, wherein the mobile device is an autonomous robot.

37. The method of claim 9, wherein the mobile device is adapted for cleaning.

38. The method of claim 31, further comprising:

moving the mobile device from the fourth pose to a charging station at a fourth location;

moving the mobile device from the charging station back to the fourth pose.

39. A robot configured to determine its location and orientation (pose) in an environment in which a signal, external to the robot, is present, the robot comprising:

a signal sensor configured to detect a property of the signal;

a movement system configured to move the robot from a first pose comprising a first location in the environment at a first time to a second pose comprising a second location at a second time, the second location proximate to the first location;

a motion sensor configured to detect a change in location between the first pose and the second pose;

a local signal estimator configured to predict the value of the signal property at a plurality of poses, the plurality of poses comprising poses with respective predefined locations proximate to the first pose; and

a localization component configured to estimate the robot'"'"'s second pose based at least in part on a value of the signal property as detected by the signal sensor at the second pose, the change in pose between the first pose and the second pose as detected by the motion sensor, and a predicted value of the signal property at the estimated second pose based at least in part on one or more of the values predicted by the local signal estimator.

40. The robot of claim 39, wherein the motion sensor is configured to detect the change in location between the first pose and the second pose by detecting a relative change in location.

41. The robot of claim 39, wherein two locations are proximate if they are less than 2 meters apart.

42. The robot of claim 41, wherein two locations are proximate if they are less than 2 feet apart.

43. The robot of claim 39, wherein:

the movement system is further configured to move the robot from the second location at the second time to a third pose comprising a third location in the environment at a third time, the third location not proximate to the first location;

the motion sensor is further configured to detect a change in location between the second location and the third location;

the local signal estimator is further configured to predict values of the signal property at a plurality of locations proximate to the third location based, at least in part, on extrapolating from previously predicted values at one or more locations proximate to both the first location and the third location; and

the localization component is further configured to estimate the robot'"'"'s third location based at least in part on a value of the signal property as detected by the signal sensor at the third location, the change in location between the second location and the third location as detected by the motion sensor, and a value of the signal property at the estimated third location as predicted by the local signal estimator.

44. The robot of claim 43 wherein the local signal estimator is further configured to introduce noise into the extrapolation.

45. The robot of claim 39, wherein the motion sensor is further configured to detect a relative change in orientation between the first pose and the second pose.

46. The robot of claim 45, wherein the localization component is further configured to estimate the robot'"'"'s second pose based at least in part on a rotational variance measure associated with how the value of the signal property detected by the signal sensor depends on the orientation of the robot.

47. A mobile device comprising:

a signal sensor configured to provide a measurement of a property of a localization signal at a time, the measurement affected at least in part by at least one of a location or an orientation (comprising a pose) of the mobile device at the time;

a motion sensor configured to measure a change in pose of the mobile device from an earlier pose to a subsequent pose;

a calibration processing component configured to calculate a correlation factor associating the localization signal property to a respective pose comprising a correlated location such that the correlation factor relates the correlated location to a value of the localization signal property measured by the signal sensor at the respective pose, wherein the correlation factor is calculated based at least in part on at least a first measurement of the property of the localization signal by the signal sensor at a first mobile device pose at a first time and a second measurement of the localization signal property by the signal sensor at a second mobile device pose at a second time;

a node initialization processing component configured to calculate expected measures of the localization signal property at each of a first set of nodes, the nodes in the first set of nodes having respective predefined poses, the expected measures based at least in part on the respective node'"'"'s location and the correlation factor; and

a localization processing component configured to estimate a current pose of the mobile device based at least in part on;

a previously estimated pose,

an expected measure of the property localization signal for the current pose calculated at least in part based on the expected measures for two or more of the nodes in the first set of nodes,

a measure of the localization signal property as reported by the signal sensor at the current pose,

the calculated rotational variation, and

a change in pose from the previously estimated pose and the current pose as measured by the motion sensor.

48. The mobile device of claim 47 wherein:

the calibration processing component is further configured to calculate a rotational variation of the signal sensor based at least in part on at least the first measurement and the second measurement; and

the node initialization processing component is further configured so that the respective expected measures are further based at least in part on the calculated rotational variation.