How to train your robot (to feed you dinner) — ScienceEach day

It's a time-consuming and infrequently awkward job, one largely finished out of necessity slightly than alternative.

Researchers on the University of Washington are engaged on a robotic system that may assist make it simpler. After figuring out totally different meals on a plate, the robot can strategize how to use a fork to decide up and ship the specified chew to an individual's mouth.

The group revealed its leads to a sequence of papers: One was lately revealed in IEEE Robotics and Automation Letters, whereas the opposite can be offered March 13 on the ACM/IEEE International Conference on Human-Robot Interaction in South Korea.

"Being dependent on a caregiver to feed every bite every day takes away a person's sense of independence," stated corresponding creator Siddhartha Srinivasa, the Boeing Endowed Professor within the UW's Paul G. Allen School of Computer Science & Engineering. "Our goal with this project is to give people a bit more control over their lives."

The thought was to develop an autonomous feeding system that may be hooked up to individuals's wheelchairs and feed individuals no matter they wished to eat.

"When we started the project we realized: There are so many ways that people can eat a piece of food depending on its size, shape or consistency. How do we start?" stated co-author Tapomayukh Bhattacharjee, a postdoctoral analysis affiliate within the Allen School. "So we set up an experiment to see how humans eat common foods like grapes and carrots."

The researchers organized plates with a few dozen totally different sorts of meals, ranging in consistency from laborious carrots to gentle bananas. The plates additionally included meals like tomatoes and grapes, which have a tricky pores and skin and gentle insides. Then the group gave volunteers a fork and requested them to decide up totally different items of meals and feed them to a model. The fork contained a sensor to measure how a lot pressure individuals used once they picked up meals.

The volunteers used numerous methods to decide up meals with totally different consistencies. For instance, individuals skewered gentle gadgets like bananas at an angle to maintain them from slipping off the fork. For gadgets like carrots and grapes, the volunteers tended to use wiggling motions to enhance the pressure and spear every chew.

"People seemed to use different strategies not just based on the size and shape of the food but also how hard or soft it is. But do we actually need to do that?" Bhattacharjee stated. "We decided to do an experiment with the robot where we had it skewer food until the fork reached a certain depth inside, regardless of the type of food."

The robot used the identical force-and-skewering technique to attempt to decide up all of the items of meals, no matter their consistency. It was ready to decide up laborious meals, but it surely struggled with gentle meals and people with robust skins and gentle insides. So robots, like people, want to alter how a lot pressure and angle they use to decide up totally different sorts of meals.

The group additionally famous that the acts of choosing up a chunk of meals and feeding it to somebody should not unbiased of one another. Volunteers typically would particularly orient a chunk of meals on the fork in order that it might be eaten simply.

"You can pick up a carrot stick by skewering it in the center of the stick, but it will be difficult for a person to eat," Bhattacharjee stated. "On the other hand, if you pick it up on one of the ends and then tilt the carrot toward someone's mouth, it's easier to take a bite."

To design a skewering and feeding technique that adjustments primarily based on the meals merchandise, the researchers mixed two totally different algorithms. First they used an object-detection algorithm known as RetinaNet, which scans the plate, identifies the kinds of meals on it and locations a body round every merchandise.

Then they developed SPNet, an algorithm that examines the kind of meals in a particular body and tells the robot the easiest way to decide up the meals. For instance, SPNet tells the robot to skewer a strawberry or a slice of banana within the center, and spear carrots at one of many two ends.

The group had the robot decide up items of meals and feed them to volunteers utilizing SPNet or a extra uniform technique: an strategy that skewered the middle of every meals merchandise no matter what it was. SPNet's various methods outperformed or carried out the identical because the uniform strategy for all of the meals.

"Many engineering challenges are not picky about their solutions, but this research is very intimately connected with people," Srinivasa stated. "If we don't take into account how easy it is for a person to take a bite, then people might not be able to use our system. There's a universe of types of food out there, so our biggest challenge is to develop strategies that can deal with all of them."

The group is at the moment working with the Taskar Center for Accessible Technology to get suggestions from caregivers and sufferers in assisted dwelling amenities on how to enhance the system to match individuals's wants.

"Ultimately our goal is for our robot to help people have their lunch or dinner on their own," Srinivasa stated. "But the point is not to replace caregivers: We want to empower them. With a robot to help, the caregiver can set up the plate, and then do something else while the person eats."