It’ll be a while before consumer robots become mainstream, according to a new study by Juniper Research. “Worldwide Consumer Robotics: Markets & Strategies 2015-2020” forecasts that more than 1 in 10 American households will own a consumer robot by the end of the decade, up from under 1 in 25 in 2015.

Shipments during the early stages of the market will be dominated by task-oriented robots, including robotic lawn mowers and vacuums that “offer tremendous ‘fire and forget’ type convenience for consumers.” Nothing really unexpected there, as iRobot’s Roomba is the most successful consumer robot of all time with more than 14 million units sold worldwide.

Juniper Research defines a consumer robot as an “autonomous, mobile electromechanical machine, capable of being programmed and re-programmed, that is used in the home or has non-commercial applications. It should be able to perceive its environment to some extent and react to it.”

The study adds that “more complex robots,” such as personal robots like Pepper, “are heavily limited by present-day technology” and that “to meet consumer expectations, smarter, more contextually aware robots are required.”

A major step in the medium-term for personal robots to get to that next level, the study says, will be a leap forward in AI that “will demand not only more computing power, but also much greater efficiency if processing is to be offloaded from the cloud.”

Other hurdles standing in the way of consumer robots include cost and trust. Component economies of scale have yet to be achieved, while R&D costs are high. Meanwhile, studies indicate that trust between robots and humans is rapidly eroding, even if a robot is able to perform better than a human on average.

‘The state of consumer robotics could be compared to the PC in the late 70s’ noted research author Steffen Sorrell. “Venture capitalist and corporate investment has ramped up tremendously recently – they know that this is the start of a paradigm shift in the way we use and interact with machines.”