The MIT Faculty has made this article openly available. Please share how this access benefits you. Your story matters.

State-of-the-art RFID localization systems fall under two categories. The first category operates with off-the-shelf narrowband RFID tags but makes restrictive assumptions on the environment or the tag's movement patterns. The second category does not make such restrictive assumptions; however, it requires designing new ultrawideband hardware for RFIDs and uses the large bandwidth to directly compute a tag's 3D location. Hence, while the first category is restrictive, the second one requires replacing the billions of RFIDs already produced and deployed annually.This paper presents RFind, a new technology that brings the benefits of ultra-wideband localization to the billions of RFIDs in today's world. RFind does not require changing today's passive narrowband RFID tags. Instead, it leverages their underlying physical properties to emulate a very large bandwidth and uses it for localization. Our empirical results demonstrate that RFind can emulate over 220MHz of bandwidth on tags designed with a communication bandwidth of only tens to hundreds of kHz, while remaining compliant with FCC regulations. This, combined with a new superresolution algorithm over this bandwidth, enables RFind to perform 3D localization with sub-centimeter accuracy in each of the x/y/z dimensions, without making any restrictive assumptions on the tag's motion or the environment.

Quantitative evaluation of the microstructural state of a specimen can be deduced from knowledge of the sample's absolute acoustic nonlinearity parameter, β, making the measurement of β a powerful tool in the NDE toolbox. However, the various methods used in the past to measure β each suffer from significant limitations. Piezoelectric contact transducers are sensitive to nonlinear signals, cheap, and simple to use, but they are hindered by the variability of the interfacial contact between transducer and specimen surface. Laser interferometry provides non-contact detection, but requires carefully prepared specimens or complicated optics to maximize sensitivity to the higher harmonic components of a received waveform. Additionally, laser interferometry is expensive and relatively difficult to use in the field. Air-coupled piezoelectric transducers offer the strengths of both of these technologies and the weaknesses of neither, but are notoriously difficult to calibrate for use in nonlinear measurements. This work proposes a hybrid modeling and experimental approach to air-coupled transducer calibration and the use of this calibration in a model-based optimization to determine the absolute β parameter of the material under investigation. This approach is applied to aluminum and fused silica, which are both well-documented materials and provide a strong reference for comparison of experimental and modeling results.

The shortage of skilled workers who can use robots is a crucial issue hampering the growth of manufacturing industries. We present a new type of workforce training system, TeachBot, in which a robotic instructor delivers a series of interactive lectures using graphics and physical demonstration of its arm movements. Furthermore, the TeachBot allows learners to physically interact with the robot. This new human-computer interface, integrating oral and graphical instructions with motion demonstration and physical touch, enables to create engaging training materials. Effective learning takes place when the learner simultaneously interacts with an embodiment of new knowledge. We apply this “Learning by Touching” methodology to teach basic concepts, e.g. how a shaft encoder and feedback control work. In a pilot randomized control test with a small number of human subjects, we find suggestive evidence that Learning by Touching enhances learning effectiveness in this robotic context for adult learners. Students whose learning experience included touching the robot as opposed to watching it delivers the lessons showed gains in their ability to integrate knowledge about robotics. The “touching” group showed statistically significant gains in self-efficacy, which is an important antecedent to further learning and successful use of new technologies, as well as gains in knowledge about robotic concepts that trend toward significance.

No abstract