Researchers:
Daisuke Sakamoto
Koichiro Honda
Masahiko Inami
Takeo Igarashi
Paper Link:
http://delivery.acm.org/10.1145/1520000/1518733/p197-sakamoto.pdf?key1=1518733&key2=0508845621&coll=&dl=&CFID=75086870&CFTOKEN=52161669
Sakamoto et al. have identified the problem that robots designed for the home are completely autonomous and lack controls for the user to specify what a robot needs to do.
So Sakamoto et al. developed what they called "Sketch and Run" for the Roomba. (a vacuum cleaner robot) The Sketch and Run system observes the robot from a system of several cameras positioned on the ceiling looking straight down; The user can then issue pen-stroke based commands on a hand-held controller (in this case, a laptop) to tell the robot where to move, vacuum, pause, resume, and
An important part of the system is that the user can issue a command and leave the robot to its work which is an important feature in any robot. Hence the name, "Sketch and Run." This principle is called asynchronism.
___________________________
My Spill:
Sketch and Run's system provides an easy-to-use way to asynchronously control robots that can easily be extended to other robot models. That's great.
I'd love to see robotics come into the home with a user friendly model.
I have a few problems with sketch and run:
First is that the command strokes for resume and stop seem too similar.
Second is that the command strokes for go home and pause seem too complex.
Third is the fact that they claim that they want to make a user friendly control system that is affordable yet they use a system of four ceiling mounted cameras in order to make their system work.
Finally, the camera system was flawed in that the Roomba would go out of the cameras detection if they went to the overlapping edges of the cameras and control would become difficult.
These kinds of serious kinks need to be worked out in the future.
They need to make simpler strokes for control and an easier way to detect the robot other than a bunch of ceiling mounted cameras.
Ross, indeed it is doubtful if the authors presented a 'simple' sketch based interface as they claimed in the starting.
ReplyDeleteNot just the complex setup required, it is also questionable if they could extend their sketch interface to robots more advanced than Roomba. On the face, it seems plausible only in controlling spatial movement of robots.
Thanks for the informative blog !
$ubodh Prabhu
http://subodhprabhu.blogspot.com/
(from CPSC 436 class)