Shape Tracking with Touch Sensors
High performance contour tracking can be achieved with simple sensing and control. Ph.D. student Liangchuan Mi has implemented a system that tracks unknown shapes with a joystick sensor mounted on an Adept SCARA robot. The figure below shows the setup, a tracked object, and the resulting contour.
The joystick’s limited force sensing is combined with the Adept’s high positional accuracy to yield precise contact measurements. Force control is carried out with a simple feedback loop to avoid unnecessary oscillations.
Due to contact friction and random sensor noise, prediction of the tangential motion from force measurements along is very unreliable, which affects the tracking speed significantly. A position control strategy is devised to update the tracking direction based on a quadratic fit to local contour data. Fitting yields more accurate shape approximation while the use of local data keeps the computational cost low for constant updates. The result is faster tracking with hardly any loss of shape accuracy. The left figure below shows the partial contour during tracking, the tangential direction (blue) predicted from force measurement alone, and the estimated direction (green). The right figure below is a quadratic fit over the recent tracking directions to estimate the direction at the currrent contact.
The next three figures compares the real shape, its contour generated with tracking speed 0.3mm/s, and its contour generated with tracking speed 5.0mm/s. The two contours approximate the original shape with almost no visual difference that they are displayed side by side instead of superposed on the shape.
To further speed up tracking, Liangchuan is planning to use stepper motors and an FPGA (Field Programmable Gate Arrays) board to control controlling joystick motions directly.
- Liangchuan Mi and Yan-Bin Jia. High precision contour tracking with a joystick sensor. In Proceedings of the IEEE/RSJ International Conference on Intelligent Robots and Systems, pp. 804-809, Sendai, Japan, Sep 28 – Oct 2, 2004.
This material is based upon work supported by an NSF CAREER Award 0133681.
Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.
Last updated on July 26. 2004