Structured English brings robots closer to everyday users

Nov 23, 2011 By Anne Ju
Mae demonstrates her understanding of structured-English commands during a demo in Rhodes Hall. Background from left, graduate students Vasu Raman, Jim Jing and Cameron Finucane stand with Hadas Kress-Gazit. Photo: Lindsay France

(PhysOrg.com) -- Move over, Jetsons. A humanoid robot named Mae is traipsing around Cornell's Autonomous Systems Lab, guided by plain-English instructions and sometimes even appearing to get frustrated.

Mae understands and executes English commands, thanks to algorithms and a software toolkit called Linear Temporal Logic Mission Planning (LTLMoP) being developed in the lab of Hadas Kress-Gazit, assistant professor of mechanical and .

According to Kress-Gazit, the future of robotics is in the ability of robots to easily understand everyday users and to act reliably in different situations.

"The big picture is that we want to have anybody tell the robot what to do," explained Kress-Gazit, who studies how to create provably correct, high-level behaviors for robots. "You don't want to have a programmer who's been doing the job forever to have to write the code for every single behavior, as is currently done in the field. You want to take what someone said and automatically generate the code for the robot to successfully accomplish its task."

The LTLMoP toolkit combines logic, language and control algorithms. The group has demonstrated the algorithms by getting Mae, a 2-foot robot NAO humanoid made by Aldebaran Robotics, to simulate looking for missing items in a grocery store while also avoiding spills in the aisles. Depending on what she finds, the robot takes action based on the specifications that were given to her.

The "store" is located in the Rhodes Hall Lab. Mae knows how to react in certain situations -- for example, if a "missing item" is encountered, she alerts a manager. If she sees a "spill," she'll avoid the area.

Traditionally, a controller for these relatively complex tasks requires specifically programming the robot to react in every conceivable state it may find itself in. This is the tedious and error-prone nature of robotics today, the researchers say. There's no guarantee that the code has accounted for every situation, and that it will work. There's also no guarantee the behavior is even possible.

For their work, the Cornell researchers are looking at how to provide explanations to the user when, for whatever reason, a task cannot be done. That kind of feedback from the robot does not exist in robotics today, Kress-Gazit says.

In LTLMoP, a high-level specification can be written in structured English. For example, Mae is told to visit all the corners of the "store" and to look side to side while walking through the aisles. The commands can be written concisely because the understands breaking the store into "regions," -- and prepositional statements, such as "between" and conditional statements like "if … then."

"Instead of giving it a list of things to do in order, you give it a specification of the sort of behavior it should exhibit at all times," said graduate student Cameron Finucane, who works on the LTLMoP platform.

Kress-Gazit's research is supported by the National Science Foundation CAREER program and a Department of Defense Multidisciplinary University Research Initiative.

Explore further: Forging a photo is easy, but how do you spot a fake?

Related Stories

Kondo Robot releases a hexapod robot kit (w/ video)

Apr 11, 2011

(PhysOrg.com) -- Kondo Robot, a Japan-based robotics company known for selling robotics kits which often end up in robot-on-robot battles, announced the release of a new robot kit. The kit, named the KMR-M6 ...

Researchers give robot ability to learn (w/ Video)

Aug 02, 2011

Researchers with the Hasegawa Group at the Tokyo Institute of Technology have created a robot that is capable of applying learned concepts to perform new tasks. Using a type of self-replicating neural technology ...

Recommended for you

Forging a photo is easy, but how do you spot a fake?

18 hours ago

Faking photographs is not a new phenomenon. The Cottingley Fairies seemed convincing to some in 1917, just as the images recently broadcast on Russian television, purporting to be satellite images showin ...

Algorithm, not live committee, performs author ranking

22 hours ago

Thousands of authors' works enter the public domain each year, but only a small number of them end up being widely available. So how to choose the ones taking center-stage? And how well can a machine-learning ...

Professor proposes alternative to 'Turing Test'

Nov 19, 2014

(Phys.org) —A Georgia Tech professor is offering an alternative to the celebrated "Turing Test" to determine whether a machine or computer program exhibits human-level intelligence. The Turing Test - originally ...

Image descriptions from computers show gains

Nov 18, 2014

"Man in black shirt is playing guitar." "Man in blue wetsuit is surfing on wave." "Black and white dog jumps over bar." The picture captions were not written by humans but through software capable of accurately ...

Converting data into knowledge

Nov 17, 2014

When a movie-streaming service recommends a new film you might like, sometimes that recommendation becomes a new favorite; other times, the computer's suggestion really misses the mark. Yisong Yue, assistant ...

User comments : 1

Adjust slider to filter visible comments by rank

Display comments: newest first

plasticpower
5 / 5 (1) Nov 23, 2011
The real question is: will it blend?

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.