TOKYO — Honda’s walking, talking interactive robot is running into glitches in its new job as a museum guide in Tokyo.
The bubbleheaded Asimo machine had problems telling the difference between people raising their hands to ask questions and those aiming their smartphones to take photos at the Miraikan science museum.
It froze midaction and repeated a programmed remark, “Who wants to ask Asimo a question?”
The robot guide, shown to reporters Wednesday, is connected by wireless to six sensors in the ceiling to discern where a crowd is gathering.
- 2 people killed in Seattle-area windstorm identified
- High winds stall firefighting efforts, fuel Tunk Block, Lime Belt fires
- Steven Hauschka's 60-yard FG gives Seahawks final edge over Chargers
- Jack Zduriencik’s M’s legacy: More than 3 dozen departed managers, coaches, scouts, staffers
- Offense needs big kick as Seahawks snag 16-15 victory
Most Read Stories
It has no voice recognition and responds to written questions selected from a touch-panel device, meaning that interacting with Asimo was even less natural or entertaining than the typical dead-end conversation with a smartphone.
Honda’s robotics technology, although among the most advanced for mobility, has come under fire as lacking practical applications and being little more than an expensive toy.
Asimo was too sensitive to go into radiated areas after the 2011 Fukushima nuclear crisis, and Honda had to do new work to develop robotic arms that could be used in reactors.
Satoshi Shigemi, who oversees Honda’s robotics technology, acknowledged that more work was needed.
He said the goal is to have Asimo recognize who is talking to it, such as an adult versus a child, and respond accordingly.
It has a long way to go.
“Right now, it can recognize a child waving to it, but it’s not able to comprehend the meaning of the waving,” he told reporters.
A possible future use for Asimo would be to help people buy tickets from vending machines at train stations, Shigemi said.
“The tasks that any child can perform, such as language interpretation, are among the most challenging for a computerized agent, whether robotic or not,” said Ingrid Zukerman, professor of information technology at Australia’s Monash University.
The effort is complex, including making out speech, then connecting it with knowledge to make out meanings, and so the state-of-the-art language interpretation is still limited to specific requests such as bus schedules and airline bookings, Zukerman said.
In Wednesday’s demonstration, the robot walked around and gesticulated as it answered a question about its features while projecting diagrams on a screen.
Honda said it was programmed to reply to about 100 questions. Four were chosen randomly and popped up on the touch panels for the audience to choose.
Honda’s humanoid prototype was first shown in 1996, and it has grown smaller and nimbler over the years.
In previous demonstrations, Asimo has shown it can run, hop on one foot and kick a soccer ball.
It has fingers with joints and can open a thermos lid and pour a drink. It has image sensors and can make its way through objects in a room.