Thesis (Ph. D.)--University of Washington, 1999. / Vita. Includes bibliographical references (leaves 100-106).

This thesis presents the development of an instrumented compliant wrist mechanism which serves as an interface between robotic platforms and their environments in order to detect surface positions and orientations. Although inspired by similar existing devices, additional features such as noncontact distance estimations, a simplified physical structure, and wireless operation were incorporated into the design. The primary role envisioned for this mechanism was for enabling robotic manipulators to perform surface following tasks prior to contact as this was one requirement of a larger project involving inspection of surfaces. The information produced by the compliant wrist system can be used to guide robotic devices in their workspace by providing real-time proximity detection and collision detection of objects. Compliance in robotic devices has attracted the attention of many researchers due to the multitude of benefits it offers. In the scope of this work, the main advantage of compliance is that it allows rigid structures to come into contact with possibly fragile objects. Combined with instrumentation for detecting the deflections produced by this compliance, closed-loop control can be achieved, increasing the number of viable applications for an initially open-loop system. Custom fabrication of a prototype device was completed to physically test operation of the designed system. The prototype incorporates a microcontroller to govern the internal operations of the device such as sensor data collection and processing. By performing many computation tasks directly on the device, robotic controllers are able to dedicate more of their time to more important tasks such as path planning and object avoidance by using the pre-conditioned compliant device data. Extensive work has also gone into the refinement of sensor signals coming from the key infrared distance measurement sensors used in the device. A calibration procedure was developed to decrease inter-sensor variability due to the method of manufacturing of these sensors. Noise reduction in the signals is achieved via a digital filtering process. The evaluation of the performance of the device is achieved through the collection of a large amount of sensor data for use in characterisation of the sensor and overall system behavior. This comes in the form of a statistical analysis of the sensor outputs to determine signal stability and accuracy. Additionally, the operation of the device is validated by its integration onto a manipulator robot and incorporating the data generated into the robot’s control loop.