Search for: All records

The personalization of therapy for children with Autism Spectrum Disorder (ASD) has been found to be crucial in comparison to a universal approach. This personalization in therapy demands the ability to adapt to the individual’s needs and engagement levels to avoid disinterest or meltdowns. This paper proposes the first step towards forecasting engagement of children with ASD during therapy sessions using Blood Volume Pulse (BVP). The BVP data is collected from an interactive session between two children with ASD in the presence of a NAO robot, and the forecast is made using a Deep Learning architecture combining Convolutional Neural Networks (CNNs) and Long-short term Memory (LSTM). Out of the three networks tested: LSTM, CNN and CNN+LSTM, the latter was found to outperform the others and gave a coefficient of determination of 0.955. The forecast was done using less than 3 minutes of prior BVP data to forecast 3 minutes into the future time steps. 

In Annual SuperCollider Conference for Kentucky NSF EPSCoR Undergraduate Research Poster Session, Louisville, Kentucky, April 22, 2022. 

In Proc. of the 20th Annual Kentucky Posters at the Capitol Undergraduate Research Poster Session, Frankfort, Kentucky, March 3, 2022. 

In Proc. of Conference of the International Society for Research on Emotion (ISRE), Los Angeles, California, July 15-18, 2022. 

Monitoring daily activities is essential for home service robots to take care of the older adults who live alone in their homes. In this article, we proposed a sound-based human activity monitoring (SoHAM) framework by recognizing sound events in a home environment. First, the method of context-aware sound event recognition (CoSER) is developed, which uses contextual information to disambiguate sound events. The locational context of sound events is estimated by fusing the data from the distributed passive infrared (PIR) sensors deployed in the home. A two-level dynamic Bayesian network (DBN) is used to model the intratemporal and intertemporal constraints between the context and the sound events. Second, dynamic sliding time window-based human action recognition (DTW-HaR) is developed to estimate active sound event segments with their labels and durations, then infer actions and their durations. Finally, a conditional random field (CRF) model is proposed to predict human activities based on the recognized action, location, and time. We conducted experiments in our robot-integrated smart home (RiSH) testbed to evaluate the proposed framework. The obtained results show the effectiveness and accuracy of CoSER, action recognition, and human activity monitoring.