Hobbit: the EU care robot that ran 372 days in 18 homes

The idea that an older person looks after the robot

Hobbit is a home service robot from an EU FP7 project led by TU Wien. The project ran from November 2011 to August 2015 at a total cost of about €3.82 million. Its core concept was “Mutual Care”: if an older person looks after the robot like a pet, the thinking went, accepting the robot’s help gets easier. The main task was fall prevention and detection, with the robot picking objects up from the floor and raising an alarm in an emergency. Lund University researcher Susanne Frennert said, “It’s a bit like R2D2, but it has a giraffe neck as well.” A Swedish report described a squarish body like a tall copier topped by a boxy head with two round eyes. The authors write that users wanted a head so they knew whom to speak to. Workshop participants modeled soft, washable, slim bodies with two arms, but only one arm was built, for cost.

372 days in 18 homes

A first prototype was shown in March 2013, after a lab study with 22 participants. In the home trials 18 older users, 7 in Austria, 4 in Greece and 7 in Sweden, lived with the robot for three weeks each. It ran 372 days in total (371 in another paper). Users were free to choose what to do with it. They found it easy and intuitive. Speech and gestures underperformed, so people relied on the touchscreen. Navigation had a service rate above 98%. The hardware cost of about €16,000 was judged too high for the target group. The battery was rated for about eight hours, but users tired of the robot after about two, and it needed at least 30% charge for safety. Thresholds up to 25 mm, missing rear sensors and heat buildup were also problems. A log-analysis paper found that gesture input looked more used than it was because of false positives.

The function they wanted most worked on only 18% of days

Users valued picking objects up from the floor most. Yet it worked without difficulty on only about 18% of the 372 days. They still wanted it, and they did not want a cheaper version without the arm. Mutual Care, by contrast, had little effect. Users did not perceive reciprocity, and emotional attachment weakened because of technical problems. The authors write, “This all indicates that reliability is a prerequisite for user acceptance.” Perceived safety against falls did not increase during the trials either. Ella, 89, who tried it for a month in Sweden, said, “It’s opened a window.” Frennert stressed that the robot complements home-care workers rather than replacing them. The authors also noted the eyes were too bright at night and suggested a fully dark face.

Reliability first, mutual care second

The home trials suggest reliability comes before the concept. A design that asks an older person to look after the robot only means something once the robot does its job dependably, as the paper implies. Hobbit was not a product, so what remains is a record of home trials instead of sales figures.

Does having people care for a robot make them accept it, or must it first stop breaking?

What we haven’t confirmed

• Whether development or commercialization continued after the project ended
• A sales price plan (sources give about €14,000 and about €16,000, with a 2013 target under €5,000)
• Results of the second home trial planned with the new robot
• Any published design award or expert design review