Robotic Sensors

The focus of our research in the area of electro-optical robotic sensors has been the development of a proximity sensor and a tactile-force sensor. Proximity sensors, measuring the distance and orientation of an object relative to the gripper, are needed in order to bridge the uncertainty gap between the gross proximity-estimation of a vision system and the direct contact required for tactile sensing.

A robust amplitude-modulated fiber-optic proximity sensor has been developed in our laboratory. The two primary characteristics of this sensor are

  • the integration of the distance and orientation transducers, and
  • active sensing.

The current transducer design of the sensor consists of one centrally placed light emitter, and eight receivers symmetrically distributed around the emitter. The transducer is connected, via fiber optics, to a new opto-electronic sensor interface designed and manufactured in the CIMLab. The interface comprises a controller, a transmitter and a receiver. The transmitter-circuit design uses a laser diode operating in the near-infrared region. The dynamic range of the interface was significantly improved via a new concept ofdynamic transmitter-intensity control.

The various calibration parameters obtained from the transducer are utilized by an active-sensing algorithm, based on the acquisition of multiple measurements at different end-effector locations. The objective of this active-sensing approach is a self-calibration process that can finetune the sensor to the surface characteristics during the gripper’s approach to the object.

The design, modeling and analysis of a novel two-layer photoelastic tactile transducer for the recovery of input force profiles was also carried in our laboratory.

The forward analysis of this transducer is carried out for both normal- and generalized-force profiles. The transducer is modeled and analyzed using closed-form equations and Finite-Element Analysis (FEA). FEA allows the analysis of a transducer having different mechanical properties in its two layers, as well as a variety of Boundary Conditions (BCs). Correspondingly, it is shown that the choice of BCs directly influences the dynamic range of the photoelastic sensor.

In the process of solving the inverse problem, that is , the recovery of the force-profile applied to the sensor, one needs to recover the phase-lead distribution from the available light-intensity distribution. A novel method was developed for this purpose. The recovery is carried out for both ideal (noise-free) and non-ideal light-intensity distributions. As part of the feasibility study for the proposed sensor, a photoelastic-transducer prototype was constructed and an optical setup was implemented with which to test it.