The Covid-19 pandemic has had a widespread effect across the globe. The major effect on health-care workers and the vulnerable populations they serve has been of particular concern. Near-complete lockdown has been a common strategy to reduce the spread of the pandemic in environments such as live-in care facilities. Robotics is a promising area of research that can assist in reducing the spread of covid-19, while also preventing the need for complete physical isolation. The research presented in this paper demonstrates a speech-controlled, self-sanitizing robot that enables the delivery of items from a visitor to a resident of a care facility. The system is automated to reduce the burden on facility staff, and it is controlled entirely through hands-free audio interaction in order to reduce transmission of the virus. We demonstrate an end-to-end delivery test, and an in-depth evaluation of the speech interface. We also recorded a speech dataset with two conditions: the talker wearing a face mask and the talker not wearing a face mask. We then used this dataset to evaluate the speech recognition system. This enabled us to test the effect of face masks on speech recognition interfaces in the context of autonomous systems.

No abstract

Recent deep-learning artificial neural networks have shown remarkable success in recognizing natural human speech, however the reasons for their success are not entirely understood. Success of these methods might be because state-of-the-art networks use recurrent layers or dilated convolutional layers that enable the network to use a time-dependent feature space. The importance of time-dependent features in human cortical mechanisms of speech perception, measured by electroencephalography (EEG) and magnetoencephalography (MEG), have also been of particular recent interest. It is possible that recurrent neural networks (RNNs) achieve their success by emulating aspects of cortical dynamics, albeit through very different computational mechanisms. In that case, we should observe commonalities in the temporal dynamics of deep-learning models, particularly in recurrent layers, and brain electrical activity (EEG) during speech perception. We explored this prediction by presenting the same sentences to both human listeners and the Deep Speech RNN and considered the temporal dynamics of the EEG and RNN units for identical sentences. We tested whether the recently discovered phenomenon of envelope phase tracking in the human EEG is also evident in RNN hidden layers. We furthermore predicted that the clustering of dissimilarity between model representations of pairs of stimuli would be similar in both RNN and EEG dynamics. We found that the dynamics of both the recurrent layer of the network and human EEG signals exhibit envelope phase tracking with similar time lags. We also computed the representational distance matrices (RDMs) of brain and network responses to speech stimuli. The model RDMs became more similar to the brain RDM when going from early network layers to later ones, and eventually peaked at the recurrent layer. These results suggest that the Deep Speech RNN captures a representation of temporal features of speech in a manner similar to human brain.

Humans make extensive use of auditory cues to interact with other humans, especially in challenging real-world acoustic environments. Multiple distinct acoustic events usually mix together in a complex auditory scene. The ability to separate and localize mixed sound in complex auditory scenes remains a demanding skill for binaural robots. In fact, binaural robots are required to disambiguate and interpret the environmental scene with only two sensors. At the same time, robots that interact with humans should be able to gain insights about the speakers in the environment, such as how many speakers are present and where they are located. For this reason, the speech signal is distinctly important among auditory stimuli commonly found in human-centered acoustic environments. In this paper, we propose a Bayesian method of selectively processing acoustic data that exploits the characteristic amplitude envelope dynamics of human speech to infer the location of speakers in the complex auditory scene. The goal was to demonstrate the effectiveness of this speech-specific temporal dynamics approach. Further, we measure how effective this method is in comparison with more traditional methods based on amplitude detection only.

One of the fundamental prerequisites for effective collaborations between interactive partners is the mutual sharing of the attentional focus on the same perceptual events. This is referred to as joint attention. In psychological, cognitive, and social sciences, its defining elements have been widely pinpointed. Also the field of human-robot interaction has extensively exploited joint attention which has been identified as a fundamental prerequisite for proficient human-robot collaborations. However, joint attention between robots and human partners is often encoded in prefixed robot behaviours that do not fully address the dynamics of interactive scenarios. We provide autonomous attentional behaviour for robotics based on a multi-sensory perception that robustly relocates the focus of attention on the same targets the human partner attends. Further, we investigated how such joint attention between a human and a robot partner improved with a new biologically-inspired memory-based attention component. We assessed the model with the humanoid robot iCub involved in performing a joint task with a human partner in a real-world unstructured scenario. The model showed a robust performance on capturing the stimulation, making a localisation decision in the right time frame, and then executing the right action. We then compared the attention performance of the robot against the human performance when stimulated from the same source across different modalities (audio-visual and audio only). The comparison showed that the model is behaving with temporal dynamics compatible with those of humans. This provides an effective solution for memory-based joint attention in real-world unstructured environments. Further, we analyzed the localisation performances (reaction time and accuracy), the results showed that the robot performed better in an audio-visual condition than an audio only condition. The performance of the robot in the audio-visual condition was relatively comparable with the behaviour of the human participants whereas it was less efficient in audio-only localisation. After a detailed analysis of the internal components of the architecture, we conclude that the differences in performance are due to egonoise which significantly affects the audio-only localisation performance.