Researchers from UCLA and Texas A&M University have designed a tool that lets users motorize everyday objects using nothing but video. For example, you could take a video of a bendable lamp being manipulated (the intended bend), the team's Robiot tool then extracts the necessary information and produces 3D models that could be 3D-printed and paired with a servomotor to motorize the lamp. No longer would users have to spend hours designing parts that might not work in their designs.
While the IoT has allowed us to interact with smart devices and platforms, automating non-connected everyday objects we interact with (lamps, trash baskets, etc.) is still far from being fully realized. The automation of physical tasks would play an essential role for those with disabilities or situational impairments that we take for granted, like opening a pantry while holding groceries, or the simple act of washing hands by turning on a faucet.
To that end, Robiot is capable of opening up the door to a new world of everyday interactive objects without the need of being familiar with CAD or other design platforms. According to the researchers, Robiot is a “design-by-demonstration tool for end-users to fabricate add-on mechanisms to actuate legacy static objects for everyday physical tasks.”
To generate the necessary mechanisms to actuate those objects, the tool extracts two essential parameters from the video demonstration. The first performs a scene and motion analysis to garner the type of joint that’s needed. It then scours a repository of 3D models looking for the object and identifies the maneuverable/ground parts, specifically where on the device a ground mechanism can be placed, and where to exert actuation.
Those two parameters enable the creation of several mechanism designs, which can then be further filtered using additional settings, including range, motion, and speed. Once finished, the tool then provides the final mechanism designs that can then be 3D-printed. The researchers tested Robiot using hour-long sessions with six participants and found they were able to design 14 mechanisms and install them onto objects using only the instructions of the platform.