Advanced Imaging

AdvancedImagingPro.com

   

Advanced Imaging Magazine

Updated: January 12th, 2011 10:01 AM CDT

Smart Software Assembles Robotic Chair

Using colored markers, blob analysis and gradient-based control algorithms, a simplified hardware architecture robustly puts chair together
The robotic chair reassembles itself
© Sandra Burns
The robotic chair reassembles itself by adding the front legs, then the chair back, and then the back legs.
Once the chair is assembled, it raises itself off the floor
© Sandra Burns
Once the chair is assembled, it raises itself off the floor.
robotic chair on pedestal and PC on right
© Sandra Burns
On the pedestal in the background is the PC that handles image processing and control tasks.
Colored markers on each component
© Nichola Feldman-Kiss
Colored markers on each component allow the vision software to easily identify the location and orientation of each piece.
Advertisement

By Kristin Lewotsky

It's surreal, yet fascinating, a seemingly ordinary kitchen chair that sits quietly on a raised platform—until with a crack it falls into pieces. There's a pause and then the chair seat, or base, begins to wheel busily around, hooking up to first one leg, then another, then attaching the chair back. Finally, it rises slowly from a splayed flat position, in a process creator Max Dean likens to that of a fawn standing for the first time.

Dean conceived of the chair back in 1985. To build it, he teamed with artist and industrial designer Matt Donovan of the Department of Mechanical and Process Engineering at ETH Zurich for the mechanical design and Professor Raffaello D'Andrea of ETH Zurich for the overall system architecture and algorithm development.

The chair sits on an 8-foot-by-8-foot-by-6-inch platform and consists of six elements: the base, the four legs, and the back. High-torque servo motors in the base propel it around and allow it to roll off any of the components it lands atop. Brackets at the top of each leg fit into sockets in each corner of the base, where they are pinned into place by additional servo motors. A small robotic arm and clamp assembly attaches to the chair back and raises it into place. Additional motors raise the fully assembled chair to an upright position. The motors operate open loop, receiving commands developed by the control system using input from a vision system.

From the beginning, the design philosophy was one of simplicity. "We decided we would have a camera up above that would transmit information to a laptop computer," says D'Andrea. "All the number crunching and algorithmic stuff would be done by the laptop and the chair would be commanded via wireless [transmission]. It's the most flexible approach; it's also easy to buy a laptop with a lot of computing power. The chair itself doesn't have to be that smart, it just has to follow commands."

The vision hardware consists of a single 1392 × 1040 progressive-scan CCD camera (Allied Vision Technologies Marlin) suspended 12 to 16 feet directly over the chair, and fitted with an appropriate objective to include the entire platform in the field of view. The 8-bit camera operates at 7.5 frames/second, with Firewire camera drivers and an integrated frame grabber.

1 2 3 4 next


Subscribe to our RSS Feeds