Robotic laundry devices and methods of use

1. A robotic system for repositioning a deformable laundry article for folding, comprising:

a conveyor disposed at a bottom of a work volume, the conveyor being configured to transfer the deformable laundry article beyond a perimeter of the work volume;

at least three lifters disposed about the perimeter of the work volume, each one of the at least three lifters comprising a gripper and at least one drive motor and configured to

move within the work volume for releasably engaging the deformable laundry article, and

operate at least one of independently of and in tandem with another of the at least three lifters to suspend the deformable laundry article above the conveyor;

two or more sensors disposed at two or more fixed locations about the work volume, each of the two or more sensors configured to

at least one of detect one or more features and capture one or more images of the deformable laundry article disposed within the work volume, and

output a signal comprising the at least one of the detected one of one or more features and captured one or more images;

a memory storing data indicative of one or more repositioned deformable laundry articles; and

a controller in operative communication with the memory, the two or more sensors, and the at least one drive motor and gripper of each of the at least three lifters, the controller configured to:

receive the output signal from one or more of the two or more sensors,

identify, based on the output signal, one or more grip points adjacent one or more free-hanging edges of the deformable laundry article suspended above the conveyor by one of the at least three lifters, the gripper associated with the one of the at least three lifters being engaged with the deformable laundry article and positioned at a suspension height,

instruct another of the at least three lifters to grip, with an associated gripper, one of the one or more grip points on the deformable laundry article and lift the gripped one of the one or more grip points to the suspension height along a shared axis defined by an associated gripper of each of the engaged one and another lifter,

determine whether the deformable laundry article suspended by the one and the another of the at least three lifters is repositioned based on a comparison of the output signal to the data indicative of one or more repositioned deformable laundry articles, and

instruct, based on a determination of the deformable laundry article being repositioned, the one and the another of the at least three lifters to

lower the repositioned deformable laundry article onto the conveyor, and

release the repositioned deformable laundry article disposed on the conveyor.

2. The system of claim 1 , wherein the controller, upon determining the laundry article is not repositioned, is further configured to

detect, based on the received one or more output signals, a grip point disposed on the deformable laundry article below the suspension height,

instruct an available gripper of the at least three lifters to engage the detected grip point,

determine the available gripper is engaged with the grip point,

instruct the engaged available gripper to lift the grip point to the suspension height, and

instruct the engaged available gripper to release the laundry article.

3. The system of claim 2 , wherein the controller is further configured to iteratively execute a loop comprising

detecting, based on the received one or more output signals, a grip point disposed on the deformable laundry article below the suspension height,

instructing an available gripper of one of the at least three lifters to engage the grip point,

determining the available gripper is engaged with the grip point,

instructing the engaged available gripper to lift the grip point to the suspension height, and

instructing a longest engaged gripper to release the laundry article, until the controller at least one of:

determines a match of the output signal to the data indicative of one or more repositioned laundry articles,

exceeds a threshold number of iterations without determining a match, and

exceeds a threshold time limit without determining a match.

4. The system of claim 1 , wherein the at least three lifters are disposed about the work volume at individually anchored positions spaced apart from at least two others of the at least three lifters by between about 30 cm to 400 cm.

5. The system of claim 4 , wherein the two or more sensors comprise at least 3 sensors, each of one of the three sensor being mounted to one or more bases of the at least lifters.

6. The system of claim 1 wherein each one of the two or more sensors is calibrated to at least one of the at least three lifters.

7. The system of claim 1 , wherein the deformable laundry article is one of a plurality of deformable laundry articles comprising two or more article types of at least one of different sizes and different shapes and each of the two or more article types comprises a longest dimension of between about 4 cm to 500 cm.

8. The system of claim 7 , wherein the data indicative of one or more repositioned deformable laundry articles comprises data associated with of each of the two or more article types.

9. The system of claim 1 , wherein the controller is configured instruct the one and the another of the at least three lifters to lower the article onto the conveyor at least one of at a rate of movement of the conveyor and along a non vertical slope.

10. The system of claim 1 , wherein the two or more sensors comprise at least one of a 3-D point cloud sensor, a 2-D camera, LIDAR, LADAR, a sonar proximity sensor, an ultrasonic ranging sensor, a radar sensor, and a pair of stereo depth cameras.

11. The system of claim 10 , wherein at least one of the two or more sensors is a 3-D point cloud sensor, the memory further comprises a neural network, and the comparison of the output signal to the data indicative of one or more repositioned deformable laundry articles comprises an analysis based on a neural network.

12. The system of claim 11 , wherein the neural network comprises a trained classifier configured to generate a descriptor indicative of trained class comprising at least one of repositioned and not repositioned.

13. The system of claim 12 , wherein the trained classifier comprises a plurality of classes trained on images of at least one of a plurality of article types and sizes of a plurality of laundry articles.

14. The system of claim 12 , wherein the analysis further comprises determining whether the comparison meets or exceeds a threshold confidence value indicative of an unfolded state.

15. The system of claim 14 ,

wherein the controller is further configured to adjust the threshold confidence value after at least one of a period of time and a number of grips by grippers associated with the at least three lifters for determining a match between the repositioned deformable laundry article and the data,

wherein the period of time is between about 5 seconds and 2 minutes, and

wherein the number of grips ranges between about 2 and 20.

16. The system of claim 1 , wherein the at least one drive motor comprises a pan motor, a tilt motor, and an extend motor configured to engage an extendable arm, the extendable arm comprising no more than one joint and terminating at an associated gripper comprising at least two actuatable fingers.

17. The system of claim 16 , wherein the extendable arm comprises a monolithic arm comprising no joints and being configured to fixedly receive thereon a wrist assembly comprising the associated gripper and at least one of an actuatable swivel joint and an actuatable hinge joint disposed between the associated gripper and the arm.

18. The system of claim 17 , wherein the actuated joints are in operable communication with the controller.

19. The system of claim 1 , wherein the controller is further configured to instruct at least one of the one and the another of the at least three lifters engaged with the deformable laundry article to rotate a shared axis between associated grippers of the engaged lifters perpendicular to a run direction of the conveyor prior to instructing the one and the another of the at least three lifters to lower the repositioned deformable laundry article onto the conveyor.

20. The system of claim 19 , wherein the controller is further configured to detect at least one of a front side and a back side of the deformable laundry article and instruct at least one of the one and the another of the at least three lifters engaged with the deformable laundry article to rotate the shared axis perpendicular to a run direction of the conveyor such that the back side of the repositioned deformable laundry article is configured to contact the conveyor.

21. The system of claim 1 , wherein the system further comprises a robotic folding device in operable communication with the controller, and wherein the conveyor, operating in a forward run direction, is configured to transfer the repositioned one of a plurality of deformable laundry articles to a robotic folding device.

22. The system of claim 21 , further comprising at least one of a transfer conveyor configured to retrieve the repositioned one of the plurality of deformable laundry articles from the conveyor and transfer the one of the plurality of deformable laundry articles to a receiving surface of robotic folding device.

23. A method of selecting a laundry article repositioning routine, comprising:

receiving at a controller comprising a memory and a neural network, one or more camera images of a laundry article suspended at a suspension height by at least one lifter within a work volume of a repositioning robot, wherein the repositioning robot comprises a floor disposed at a bottom of the work volume and three or more lifters disposed about the floor at fixed positions not traced along a single straight line;

processing the received one or more camera images with a neural network trained classifier,

receiving a descriptor output from the trained classifier, the descriptor being indicative of a next action by the repositioning robot and the next action being one of a plurality of trained classes; and

instructing, based on the descriptor, one or more motor drives associated with pan, tilt, and extend motors of each of the lifters of the three or more lifters to execute the received next action.

24. The method of claim 23 , wherein the plurality of classes are manually tagged on training data comprising a plurality of images of known classes of articles, the images being ascertained by stationary sensors collocated with the three or more lifters and comprising overlapping fields of view trained on the work volume.

25. The method of claim 24 , wherein the plurality of images of the training data comprises at least one of 2-D camera images and 3-D point cloud sensor images.

26. The method of claim 23 , wherein, the next action comprises at least one of continue executing an iterative repositioning sequence for a not flat article, sweeping the repositioned flat article onto the floor, executing a recovery sequence for a missed grab, inverting a flat but inverted article, executing a shirt fold with sleeves gripped, executing a shirt fold with one sleeve and one front corner gripped, retrieving an article on the floor, and executing a shake.

27. The method of claim 26 , wherein the iterative repositioning sequence iteratively comprises an available gripper of the at least three lifters grasping and hoisting a low point of the laundry article to the suspension height of one or more grippers of one or more engaged lifters of the three or more lifters, releasing the article from a longest engaged gripper of three engaged grippers, and rotating a shared axis of a remaining two engaged grippers.