Robots Learn to Push Heavy Objects With Their Bodies, Just Like You

Whole-body shoving allows robots to move some heavy stuff

2 min read
Robots Learn to Push Heavy Objects With Their Bodies, Just Like You
Image: University of Tokyo/JSK Laboratory

The payload of a robot is a well-defined number that usually refers to how much mass its actuators or mobility system can comfortably support. The payload of a human works in a similar way, except that sometimes we can cheat, by offloading the mass of an object to the ground, and moving it purely by overcoming friction and shoving it along. For very heavy objects, doing this involves using the weight and stability of our whole bodies as well as our muscles, and robots are learning to do this, too.

img

Image: University of Tokyo/JSK Laboratory

At ICRA 2015 last week, researchers from University of Tokyo’s JSK Laboratory led by Professors Masayuki Inaba and Kei Okada presented a paper on “whole-body pushing manipulation with contact posture planning.” Or, shoving things. For humans, shoving things can be somewhat complicated, because there are a lot of different ways that a heavy object (like a big crate) can be shoved. You can put your shoulder against it and shove, put your hip against it and shove, or if it’s really heavy, lean against it with your back and shove with your legs.

The specific posture that you choose to shove something depends on what that thing weighs and how friction-y the bottom of it is, which are things that you usually don’t know before you start shoving. So what do you do? Well, you probably try shoving with your hands first, and then if it doesn’t move, you switch to a posture that allows you to exert more force.

img

Image: University of Tokyo/JSK Laboratory

The robot (an HRP-2) does the same thing: it pre-computes several different pushing postures that exert increasing amounts of force, and if it doesn’t sense the thing moving, it’ll autonomously try new strategies until it’s successful:

You’ll notice that the robot is pretty good at not falling over. This is autonomous also; it modifies its footsteps to be larger or smaller based on how far the object moves, which is detects through how much its body is tilting. 

Next, the researchers plan to “apply the proposed method to other tasks with whole-body contact,” and your guess is as good as mine as to what those tasks are going to be.

“Whole-Body Pushing Manipulation With Contact Posture Planning of Large and Heavy Object for Humanoid Robot,” by Masaki Murooka, Shunichi Nozawa, Yohei Kakiuchi, Kei Okada, and Masayuki Inaba from the University of Tokyo, was presented at ICRA 2015 in Seattle, Wash.

The Conversation (0)

How the U.S. Army Is Turning Robots Into Team Players

Engineers battle the limits of deep learning for battlefield bots

11 min read
Robot with threads near a fallen branch

RoMan, the Army Research Laboratory's robotic manipulator, considers the best way to grasp and move a tree branch at the Adelphi Laboratory Center, in Maryland.

Evan Ackerman
LightGreen

“I should probably not be standing this close," I think to myself, as the robot slowly approaches a large tree branch on the floor in front of me. It's not the size of the branch that makes me nervous—it's that the robot is operating autonomously, and that while I know what it's supposed to do, I'm not entirely sure what it will do. If everything works the way the roboticists at the U.S. Army Research Laboratory (ARL) in Adelphi, Md., expect, the robot will identify the branch, grasp it, and drag it out of the way. These folks know what they're doing, but I've spent enough time around robots that I take a small step backwards anyway.

This article is part of our special report on AI, “The Great AI Reckoning.”

The robot, named RoMan, for Robotic Manipulator, is about the size of a large lawn mower, with a tracked base that helps it handle most kinds of terrain. At the front, it has a squat torso equipped with cameras and depth sensors, as well as a pair of arms that were harvested from a prototype disaster-response robot originally developed at NASA's Jet Propulsion Laboratory for a DARPA robotics competition. RoMan's job today is roadway clearing, a multistep task that ARL wants the robot to complete as autonomously as possible. Instead of instructing the robot to grasp specific objects in specific ways and move them to specific places, the operators tell RoMan to "go clear a path." It's then up to the robot to make all the decisions necessary to achieve that objective.

Keep Reading ↓ Show less