Authors: Brandi House, Jonathan Malkin, Jeff Bilmes
Summary: In the past, those that were severely physically impaired had few options for assistance in day to day activities. Normally they would have to rely on the assistance of others, but this introduces an inconvenient dependency on others. Voicebot seeks to improve the independence of individuals with severe motor impairment by allowing them to use their voice to control a robotic arm.
Joint Movement Studies
Three methods were used to control the arm: the Forward Kinematic Model, the Inverse Kinematic Model, and the Hybrid Model. The FK model had the user control each joint one at a time. This was really easy to implement because user input is directly sent to the robotic arm. The IK model has the user just control the gripper(called the effector) and the joints move accordingly. The Hybrid model, however, uses IK to move the shoulder and elbow and then FK to move the wrist. Since any point can be reached using only two of the three joints, they had the shoulder and elbow joint move using IK and the wrist joint move using FK.
A test was done to move a ball using a simulated 2D arm to 4 sequential locations. Apparently 3/5 of the users did better with IK and 2/5 of the users did better with FK. None seemed to prefer the Hybrid model.
The findings were extended to a 3D robotic arm. Users would use the "ck" sound to switch between Position mode(gross detail) and Orientation mode(fine detail). The "ch" sound closed or opened the gripper. The robotic arm had 5 degrees of freedom(shoulder rotation,shoulder bend, elbow bend, wrist rotate, and wrist
bend.) Only three can be changed at a time so movement was separated into Position mode(gross detail) and Orientation mode(fine detail). In Position Mode the shoulder movements and elbow bend were activated and in Orientation mode the wrist was activated.
This time the arm was tested with users unfamiliar with the device. They were taught how to control it and the results are as follows:
All the users were able to finish but several complained about how difficult it was. Changing by pitch was significantly difficult and there were several false positives for the "ch" or "ck" sound.
paper
Discussion:
I guess this seems like a cool novel idea, but it seems like it still has a long way to go to be practical. As I was reading it seemed like there would be better ways to emulate arm movement that through vocal stimulation of a robotic arm.
Yeah, this does not seem like a practical idea even if it were robust. Especially with prosthetic limbs like this http://www.jhuapl.edu/newscenter/pressreleases/2007/070426.asp.
ReplyDeleteI was thinking that this might have some novel applications though for maybe a kid's display at someplace like Epcot or a Museum.