Michael Jenkin

We present DIJIT, a novel binocular robotic head expressly designed for mobile agents that behave as active observers. DIJIT's unique breadth of functionality enables active vision research and the study of human-like eye and head-neck motions, their interrelationships, and how each contributes to visual ability. DIJIT is also being used to explore the differences between how human vision employs eye/head movements to solve visual tasks and current computer vision methods. DIJIT's design features nine mechanical degrees of freedom, while the cameras and lenses provide an additional four optical degrees of freedom. The ranges and speeds of the mechanical design are comparable to human performance. DIJIT attains 85% of the peak human saccade speed. Our design includes the ranges of motion required for convergent stereo, namely, vergence, version, and cyclotorsion. Here, we present DIJIT and some aspects of its performance. We also present a novel method for saccadic camera movements, using a direct relationship between camera orientation and motor values. The resulting saccadic camera movements are close to human movements in terms of their accuracy, with 1.17$^\circ$ and 1.14$^\circ$ mean error for the left and right cameras, respectively.

The remote operation of underwater vehicles at depth is complicated by the presence of invisible and unpredictable environmental disturbances such as cross‐currents. Communicating the presence of these disturbances to an operator on the surface is made more difficult by the nature of the disturbances and the lack of visible features to highlight in the visual display presented to the operator. Here we explore the use of a novel interactive soft haptic touchpad that utilizes vibration and particle jamming to provide information about the presence and direction of cross‐currents to the operator of an ROV (remotely operated vehicle). An in‐water experiment using a thruster‐based ROV and artificially generated cross‐current was performed with nonexpert ROV operators to evaluate the effectiveness of multimodal haptic feedback to communicate complex environmental information during high‐risk operations. Advanced haptic displays can signal both the presence of external factors as well as their direction, information that can enhance operational performance as well as reduce operator cognitive load. Using haptic feedback resulted in a statistically significant reduction in cognitive load of 24.3% and an increase in positioning accuracy of 28.3% for novice operators. Deviation from an ideal path was also reduced by 29.5% for experienced operators when using haptic feedback compared to without. While this experiment took place in controlled conditions with a fixed direction cross‐current and haptic interface, this approach could be extended to communicate real‐time environmental information in real‐world unstructured environments.

Visual-tactile sensing affords abundant capabilities for contact-rich object manipulation tasks including grasping and placing. Here we introduce a shape-from-texture inspired contact shape estimation approach for visual-tactile sensors equipped with visually distinct membrane markers. Under a perspective projection camera model, measurements related to the change in marker separation upon contact are used to recover surface shape. Our approach allows for shape sensing in real time, without requiring network training or complex assumptions related to lighting, sensor geometry or marker placement. Experiments show that the surface contact shape recovered is qualitatively and quantitatively consistent with those obtained through the use of photometric stereo, the current state of the art for shape recovery in visual-tactile sensors. Importantly, our approach is applicable to a large family of sensors not equipped with photometric stereo hardware, and also to those with semi-transparent membranes. The recovery of surface shape affords new capabilities to these sensors for robotic applications, such as the estimation of contact and slippage in object manipulation tasks[1] and the use of force matching for kinesthetic teaching using multimodal visual-tactile sensing[2].

Short-term load forecasting (STLF) for residential households has become of critical importance for the secure operation of power grids as well as home energy management systems. While machine learning is effective for residential STLF, data and resource limitations hinder individual household predictions operated on local devices. In contrast, utility companies have access to broader sets of data as well as to better computational resources, and thus have the potential to deploy complex forecasting models such as Graph neural network-based models to explore the spatial-temporal relationships between households for achieving impressive STLF performance. In this work, we propose an efficient and privacy-conservative knowledge distillation-based STLF framework. This framework can improve the STLF forecasting accuracy of lightweight individual household forecasting models via leveraging the benefits of knowledge distillation and graph neural networks (GNN). Specifically, we distill the knowledge learned from a GNN model pre-trained on utility data sets into individual models without the need to access data sets of other households. Extensive experiments on real-world residential electric load datasets demonstrate the effectiveness of the proposed method.

Self-motion perception is a multi-sensory process that involves visual, vestibular, and other cues. When perception of self-motion is induced using only visual motion, vestibular cues indicate that the body remains stationary, which may bias an observer’s perception. When lowering the precision of the vestibular cue by for example, lying down or by adapting to microgravity, these biases may decrease, accompanied by a decrease in precision. To test this hypothesis, we used a move-to-target task in virtual reality. Astronauts and Earth-based controls were shown a target at a range of simulated distances. After the target disappeared, forward self-motion was induced by optic flow. Participants indicated when they thought they had arrived at the target’s previously seen location. Astronauts completed the task on Earth (supine and sitting upright) prior to space travel, early and late in space, and early and late after landing. Controls completed the experiment on Earth using a similar regime with a supine posture used to simulate being in space. While variability was similar across all conditions, the supine posture led to significantly higher gains (target distance/perceived travel distance) than the sitting posture for the astronauts pre-flight and early post-flight but not late post-flight. No difference was detected between the astronauts’ performance on Earth and onboard the ISS, indicating that judgments of traveled distance were largely unaffected by long-term exposure to microgravity. Overall, this constitutes mixed evidence as to whether non-visual cues to travel distance are integrated with relevant visual cues when self-motion is simulated using optic flow alone.

Understanding and predicting the emotional trajectory in multi-party multi-turn conversations is of great significance. Such information can be used, for example, to generate empathetic response in human-machine interaction or to inform models of pre-emptive toxicity detection. In this work, we introduce the novel problem of Predicting Emotions in Conversations (PEC) for the next turn (n+1), given combinations of textual and/or emotion input up to turn n. We systematically approach the problem by modeling three dimensions inherently connected to evoked emotions in dialogues, including (i) sequence modeling, (ii) self-dependency modeling, and (iii) recency modeling. These modeling dimensions are then incorporated into two deep neural network architectures, a sequence model and a graph convolutional network model. The former is designed to capture the sequence of utterances in a dialogue, while the latter captures the sequence of utterances and the network formation of multi-party dialogues. We perform a comprehensive empirical evaluation of the various proposed models for addressing the PEC problem. The results indicate (i) the importance of the self-dependency and recency model dimensions for the prediction task, (ii) the quality of simpler sequence models in short dialogues, (iii) the importance of the graph neural models in improving the predictions in long dialogues.

The perceptual upright results from the multisensory integration of the directions indicated by vision and gravity as well as a prior assumption that upright is towards the head. The direction of gravity is signalled by multiple cues, the predominant of which are the otoliths of the vestibular system and somatosensory information from contact with the support surface. Here, we used neutral buoyancy to remove somatosensory information while retaining vestibular cues, thus “splitting the gravity vector” leaving only the vestibular component. In this way, neutral buoyancy can be used as a microgravity analogue. We assessed spatial orientation using the oriented character recognition test (OChaRT, which yields the perceptual upright, PU) under both neutrally buoyant and terrestrial conditions. The effect of visual cues to upright (the visual effect) was reduced under neutral buoyancy compared to on land but the influence of gravity was unaffected. We found no significant change in the relative weighting of vision, gravity, or body cues, in contrast to results found both in long-duration microgravity and during head-down bed rest. These results indicate a relatively minor role for somatosensation in determining the perceptual upright in the presence of vestibular cues. Short-duration neutral buoyancy is a weak analogue for microgravity exposure in terms of its perceptual consequences compared to long-duration head-down bed rest.

Online conversations are particularly susceptible to derailment, which can manifest itself in the form of toxic communication patterns like disrespectful comments or verbal abuse. Forecasting conversation derailment predicts signs of derailment in advance enabling proactive moderation of conversations. Current state-of-the-art approaches to address this problem rely on sequence models that treat dialogues as text streams. We propose a novel model based on a graph convolutional neural network that considers dialogue user dynamics and the influence of public perception on conversation utterances. Through empirical evaluation, we show that our model effectively captures conversation dynamics and outperforms the state-of-the-art models on the CGA and CMV benchmark datasets by 1.5\% and 1.7\%, respectively.