Mohammad Nasser Saadatzi

The Use of an Autonomous Pedagogical Agent and Automatic Speech Recognition for Teaching Sight Words to Students With Autism Spectrum Disorder

In the current study, we examined the effects of an instructional package comprised of an autonomous pedagogical agent, automatic speech recognition, and constant time delay during the instruction of reading sight words aloud to young adults with autism spectrum disorder. We used a concurrent multiple baseline across participants design to evaluate the efficacy of intervention and conducted post-treatment probes to assess maintenance and generalization. Our findings suggest that all three participants acquired and maintained new sight words and demonstrated generalized responding.

Small-Group Technology-Assisted Instruction: Virtual Teacher and Robot Peer for Individuals with Autism Spectrum Disorder

The authors combined virtual reality technology and social robotics to develop a tutoring system that resembled a small-group arrangement. This tutoring system featured a virtual teacher instructing sight words, and included a humanoid robot emulating a peer. The authors used a multiple-probe design across word sets to evaluate the effects of the instructional package on the explicit acquisition and vicarious learning of sight words instructed to three children with autism spectrum disorder (ASD) and the robot peer. Results indicated that participants acquired, maintained, and generalized 100% of the words explicitly instructed to them, made fewer errors while learning the words common between them and the robot peer, and vicariously learned 94% of the words solely instructed to the robot.

Effects of a Robot Peer on the Acquisition and Observational Learning of Sight Words in Young Adults With Autism Spectrum Disorder

The authors of the current investigation developed and evaluated the effects of a tutoring system based on a small-group arrangement to two young adults with autism spectrum disorder on the acquisition, maintenance, and generalization of sight words. The tutoring system was comprised of a virtual teacher to instruct sight words, and a humanoid robot which adopted a peer metaphor, where its function was to act as an emulated peer. With the introduction of the robot peer (RP), the traditional dyadic interaction in tutoring systems was augmented to a novel triadic interaction in order to enrich the social content of the learning environment and to facilitate observational learning (OL). The virtual teacher implemented a constant time delay strategy to instruct three types of sight words: (a) target words exclusive to the participant, (b) target words common between the participant and the RP, and (c) nontarget words exclusive to the RP. In order to examine the efficacy of intervention, a multiple-probe design across three word sets, replicated across two participants, was utilized. Results indicated that both participants acquired, generalized, and maintained target words with 100% accuracy. Furthermore, the participants made fewer errors and required less instruction time to learn the words common between the participants and the RP. Finally, the participants acquired, through OL, the majority of words taught exclusively to the RP.

Modeling of a 3D acoustoelastic metamaterial energy harvester

Energy harvesters primarily depend on on a groups of unit cells to harvest energy at broadband frequencies so that each unit cell is responsible to harvest energy at a distinct frequency. Other design complexity, space, and financial profusion are required for transferring from unit-frequency to multi-frequency energy scavenging. Also, it is very unlikely to obtain expected power output if the available vibration source doesn’t match the designed loading condition (usually, unidirectional) of the device and requires rearrangement of the base structure to have projected output. In this paper we model the unique feature of acoustic metamaterial (AM), which is not only able to harvest energy at multiple frequencies using only a unit cell device, but also able to harvest energy under a variety of uncoupled (unidirectional) and coupled (multi-directional) vibration environments with an identical base structure arrangement.

Optimal accelerometer placement on a robot arm for pose estimation

The performance of robots to carry out tasks depends in part on the sensor information they can utilize. Usually, robots are fitted with angle joint encoders that are used to estimate the position and orientation (or the pose) of its end-effector. However, there are numerous situations, such as in legged locomotion, mobile manipulation, or prosthetics, where such joint sensors may not be present at every, or any joint. In this paper we study the use of inertial sensors, in particular accelerometers, placed on the robot that can be used to estimate the robot pose. Studying accelerometer placement on a robot involves many parameters that affect the performance of the intended positioning task. Parameters such as the number of accelerometers, their size, geometric placement and Signal-to-Noise Ratio (SNR) are included in our study of their effects for robot pose estimation. Due to the ubiquitous availability of inexpensive accelerometers, we investigated pose estimation gains resulting from using increasingly large numbers of sensors. Monte-Carlo simulations are performed with a two-link robot arm to obtain the expected value of an estimation error metric for different accelerometer configurations, which are then compared for optimization. Results show that, with a fixed SNR model, the pose estimation error decreases with increasing number of accelerometers, whereas for a SNR model that scales inversely to the accelerometer footprint, the pose estimation error increases with the number of accelerometers. It is also shown that the optimal placement of the accelerometers depends on the method used for pose estimation. The findings suggest that an integration-based method favors placement of accelerometers at the extremities of the robot links, whereas a kinematic-constraints-based method favors a more uniformly distributed placement along the robot links.

Characterization of large-area pressure sensitive robot skin

Sensorized robot skin has considerable promise to enhance robots’ tactile perception of surrounding environments. For physical human-robot interaction (pHRI) or autonomous manipulation, a high spatial sensor density is required, typically driven by the skin location on the robot. In our previous study, a 4x4 flexible array of strain sensors were printed and packaged onto Kapton sheets and silicone encapsulants. In this paper, we are extending the surface area of the patch to larger arrays with up to 128 tactel elements. To address scalability, sensitivity, and calibration challenges, a novel electronic module, free of the traditional signal conditioning circuitry was created. The electronic design relies on a software-based calibration scheme using high-resolution analog-to-digital converters with internal programmable gain amplifiers. In this paper, we first show the efficacy of the proposed method with a 4x4 skin array using controlled pressure tests, and then perform procedures to evaluate each sensor’s characteristics such as dynamic force-to-strain property, repeatability, and signal-to-noise-ratio. In order to handle larger sensor surfaces, an automated force-controlled test cycle was carried out. Results demonstrate that our approach leads to reliable and efficient methods for extracting tactile models for use in future interaction with collaborative robots.

Parametric investigation of scalable tactile sensors

In the near future, robots and humans will share the same environment and perform tasks cooperatively. For intuitive, safe, and reliable physical human-robot interaction (pHRI), sensorized robot skins for tactile measurements of contact are necessary. In a previous study, we presented skins consisting of strain gauge arrays encased in silicone encapsulants. Although these structures could measure normal forces applied directly onto the sensing elements, they also exhibited blind spots and response asymmetry to certain loading patterns. This study presents a parametric investigation of piezoresistive polymeric strain gauge that exhibits a symmetric omniaxial response thanks to its novel star-shaped structure. This strain gauge relies on the use of gold micro-patterned star-shaped structures with a thin layer of PEDOT:PSS which is a flexible polymer with piezoresistive properties. In this paper, the sensor is first modeled and comprehensively analyzed in the finite-element simulation environment COMSOL. Simulations include stress-strain loading for a variety of structure parameters such as gauge lengths, widths, and spacing, as well as multiple load locations relative to the gauge. Subsequently, sensors with optimized configurations obtained through simulations were fabricated using cleanroom photolithographic and spin-coating processes, and then experimentally tested. Results show a trend-wise agreement between experiments and simulations.

Fabrication of strain gauge based sensors for tactile skins

Fabricating cost effective, reliable and functional sensors for electronic skins has been a challenging undertaking for the last several decades. Application of such skins include haptic interfaces, robotic manipulation, and physical human-robot interaction. Much of our recent work has focused on producing compliant sensors that can be easily formed around objects to sense normal, tension, or shear forces. Our past designs have involved the use of flexible sensors and interconnects fabricated on Kapton substrates, and piezoresistive inks that are 3D printed using Electro Hydro Dynamic (EHD) jetting onto interdigitated electrode (IDE) structures. However, EHD print heads require a specialized nozzle and the application of a high-voltage electric field; for which, tuning process parameters can be difficult based on the choice of inks and substrates. Therefore, in this paper we explore sensor fabrication techniques using a novel wet lift-off photolithographic technique for patterning the base polymer piezoresistive material, specifically Poly(3,4-ethylenedioxythiophene)-poly(styrenesulfonate) or PEDOT:PSS. Fabricated sensors are electrically and thermally characterized, and temperaturecompensated designs are proposed and validated. Packaging techniques for sensors in polymer encapsulants are proposed and demonstrated to produce a tactile interface device for a robot.

Experimental setup for evaluating an adaptive user interface for teleoperation control

A vital part of human interactions with a machine is the control interface, which single-handedly could define the user satisfaction and the efficiency of performing a task. This paper elaborates the implementation of an experimental setup to study an adaptive algorithm that can help the user better tele-operate the robot. The formulation of the adaptive interface and associate learning algorithms are general enough to apply when the mapping between the user controls and the robot actuators is complex and/or ambiguous. The method uses a genetic algorithm to find the optimal parameters that produce the input-output mapping for teleoperation control. In this paper, we describe the experimental setup and associated results that was used to validate the adaptive interface to a differential drive robot from two different input devices; a joystick, and a Myo gesture control armband. Results show that after the learning phase, the interface converges to an intuitive mapping that can help even inexperienced users drive the system to a goal location.

An electro-dynamic 3-dimensional vibration test bed for engineering testing

Primary objective of the work is to design, fabrication and testing of a 3-dimensional Mechanical vibration test bed. Vibration testing of engineering prototype devices in mechanical and industrial laboratories is essential to understand the response of the envisioned model under physical excitation conditions. Typically, two sorts of vibration sources are available in physical environment, acoustical and mechanical. Traditionally, test bed to simulate unidirectional acoustic or mechanical vibration is used in engineering laboratories. However, a device may encounter multiple uncoupled and/or coupled loading conditions. Hence, a comprehensive test bed in essential that can simulate all possible sorts of vibration conditions. In this article, an electrodynamic vibration exciter is presented which is capable of simulating 3-dimensional uncoupled (unidirectional) and coupled excitation, in mechanical environments. The proposed model consists of three electromagnetic shakers (for mechanical excitation). A robust electrical control circuit is designed to regulate the components of the test bed through a self-developed Graphical User Interface. Finally, performance of the test bed is tested and validated using commercially available piezoelectric sensors.