CHANNELS

Aerospace

Automotive

Automated & Autonomous Vehicles

Battery & Electrification Technology

Defense

Drone TV

Electronics

Manufacturing & Prototyping

Materials

Medical

Motion Control

RF & Microwave Technology

Robotics & Automation

Sensors & IoT

Test & Measurement

Aerospace
Automotive
Automated & Autonomous Vehicles
Battery & Electrification Technology
Defense
Drone TV
Electronics
Manufacturing & Prototyping
Materials
Medical
Motion Control
RF & Microwave Technology
Robotics & Automation
Sensors & IoT
Test & Measurement
Aerospace & Defense
Search Results

A Soft Robot to Pack Your Groceries

Watch this video to learn more about a soft robotic hand that combines vision, motor-based proprioception, and soft tactile sensors to identify, sort, and pack a stream of unknown objects. A team from MIT CSAIL demonstrates its effectiveness through a grocery packing scenario.

Topics:
Automation Motion Control Robotics

More From SAE Media Group

    Transcript

    00:00:01 (logo whooshes) (bright subtle music) - Imagine showing up at the grocery store and not having to wait in line. Imagine a team of helpful robots working side-by-side with you, to help you pick up and pack all the groceries that you need for the day. This is the kind of future that we are enabling here at CSAIL,

    00:00:25 and we're developing a project we call Grocery Packing Robot. And in this project, we have a robot with very unique manipulator capabilities, that is able to pick up all sorts of items in the grocery store. Traditional robot manipulation solutions require that robots have an accurate description of the geometry of the objects that the robots have to manipulate,

    00:00:55 as well as precise location. - A lot of the previous work in bin packing has been focused on making good use of the bin space, for example, minimizing the space that is unoccupied by items. Others are more focused on speed, for example. What we do here instead is, we're trying to focus on a metric that people actually care about when you ask them, "What is important for grocery packing?",

    00:01:19 and that is that delicate items, such as chips or grapes, or other fruit are not crushed beneath heavier items. - Our solution is enabled by a unique design of the soft robotic gripper and fingers, along with several different sensors that get integrated in the robotic fingers, as well as in the environment. - First, we have an RGB depth camera that looks at a stream of unknown objects come down the conveyor belt.

    00:01:49 Our RGBD camera looks at these items, estimates its size and position, and then figures out a way where we should grasp these objects. Then we have tactile sensors on the finger pads of our gripper that can basically determine whether the item we're grasping is delicate or not, through measuring how much these items deform as soon as we grasp them.

    00:02:10 - And then we have a third modality that determines the post-configuration of the gripper, and the post-configuration of the gripper is adaptive to the shape of the object that it tries to pick up. - We pack the non-delicate items first, but put the more delicate items into a buffer to pack them later. So if we have, for example, a bunch of grapes, and a heavy tomato soup can coming down the conveyor belt, the gripper would try to grasp the bunch of grapes,

    00:02:39 and then realize that they're very delicate, so put them in the in the buffer, and then when a tomato soup can comes down the line, it would detect it as rigid, put it into the bin first, and then take the grapes from the buffer and put them into the bin afterwards, so the grapes are not crushed by the tomato soup can. - What are the key innovations that have enabled this extraordinary performance? - So first, our gripper has a special geometry

    00:03:04 that lets us deal with these items in a very delicate fashion. Also, we have new sensors that let us estimate whether items are delicate or not, and also, we have come up with this new algorithm that assigns these delicacy scores, and then sorts these items, basically implicitly by delicacy, so we can pack the groceries in the right order. - Our grocery packing solution, of course,

    00:03:27 is a research-grade prototype. We have a small robot, we have a small conveyor belt, and we have only a small set of groceries that we are able to handle within the confines of our lab. So in order to imagine a deployable solution of this idea, we have to scale up everything. We have to scale up our grasping algorithms, and this requires that we consider multiple grasping modalities,

    00:03:57 and potentially, a wider suite of sensors. We also have to consider the real-time performance of these systems, so that they can be available 24/7. Finally, the robot we use in our grocery packing system is a fixed manipulator robot. At the grocery store, you probably want mobile manipulators. - Even though this project is still in a research stage, we think it's a significant step towards showing

    00:04:27 that robots can achieve complex tasks in real environments. Maybe one day, you see one of these systems at your local grocery store. (bright subtle music)