Two-legged robot able to run without ZMP control (w/ Video)

September 3, 2014 by Bob Yirka report

( —A team of researchers at the University of Tokyo has built a small two-legged robot (named Achires) capable of running without using Zero Moment Control (ZMP)—instead it relies on high speed cameras and actuators. The team, led by Masatoshi Ishikawa, is the same one that made news recently by building a robot that could beat any human at rock-paper-scissors.

Because of balance issues, robots with two legs, rather than four, are much harder to program. Up till now, most approaches have centered around ZMP, which requires so much processing that the robots aren't able to run very fast. By eliminating the need for ZMP, the robot built in Ishikawa's laboratory, is able to run faster than any other two-legged robot thus far—4.2km/h. It's small though, with legs just 14cm long. The team notes that if torque were used to build a human-sized version of the robot, it would be able to run at speeds up to 20km/h. That would take a lot of energy of course, the diminutive version is only able to run for approximately ten seconds currently—engineers in the lab are working on increasing that time.

Achires runs by leaning forward, taking advantage of gravity, with its legs taking six steps per second. A takes snapshots (at 600fps) of the robot in motion and feeds that information to a computer which uses what it receives to control the balance of the robot. The high speed actuators allow for a speedy response to commands from the computer. The result is a robot that is able to run with a smooth gait, using far less computational muscle. The team reports that the robot is also able to perform a somersault, a stunt that is meant to show that the team plans to give the multiple capabilities that might be useful in a number of applications.

The goal of the researchers working at Ishikawa's lab is to build speed into robots, making them able to perform tasks faster than human beings, which could translate to lowered costs on production lines or when used in building or engineering projects. The idea is to look at new ways of controlling robots, sometimes using unorthodox methods, to perform tasks that we may not yet have imagined.

This video is not supported by your browser at this time.

Explore further: COMAN: Researchers build first fully-limbed passive compliant robot (w/ video)

More information:

Related Stories

A robot that runs like a cat (w/ Video)

June 17, 2013

Thanks to its legs, whose design faithfully reproduces feline morphology, EPFL's 4-legged 'cheetah-cub robot' has the same advantages as its model: It is small, light and fast.

Amber 2 robot walks with a human gait (w/ Video)

October 25, 2013

( —The engineering team at Texas A&M's Amber robotics labs has been hard at work trying to improve one area of robotics that others seem to be ignoring—getting a robot to mimic the natural gait of a human being. ...

OutRunner robot steered to fast track with Kickstarter

May 12, 2014

A startup in Pensacola, Florida, Robotics Unlimited, is turning to crowdfunding for their remote controlled fast robot called OutRunner. Running along, their robot can move up to 20 mph over different kinds of terrain, from ...

Recommended for you

Toyota promises better mileage and ride with Prius hybrid

October 13, 2015

Toyota Motor Corp. released details for its fourth-generation Prius on Tuesday, promising that improvements in the battery, engine, wind resistance and weight mean better mileage for the world's top-selling hybrid car.


Adjust slider to filter visible comments by rank

Display comments: newest first

1 / 5 (1) Sep 03, 2014
a small step in to a wrong direction
not rated yet Sep 03, 2014
Yup. While a novel idea, it requires an expensive camera and ping pong balls, plus at least how I took it, a very powerful computer (desktop) compared to any embedded processors.

Autonomy is needed, not external sensors and control loops.
5 / 5 (2) Sep 03, 2014
While a novel idea, it requires an expensive camera

It doesn't require a camera - that's just a very quick and dirty way to track where the limbs are going without going to the expense of building encoders and balance sensors into what is clearly an experimental prototype.

To make it autonomous, it would also need a body and a head with some sort of eyes to track the position relative to ground.
5 / 5 (1) Sep 04, 2014
a small step in to a wrong direction

Look at it this way: You could track and control n autonomous units by using only one central hub instead of having to equip each one with the intelligence (and sensonrs) to do so.

So you save n times the cost of extra hardware at the cost of installing one good tracking system and control unit. The larger the swarm the more this is worth it.

Please sign in to add a comment. Registration is free, and takes less than a minute. Read more

Click here to reset your password.
Sign in to get notified via email when new comments are made.