Joost C. Dessing
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Joost C. Dessing has a total of 20 co-authors. Here are the top co-authors by number of shared publications:
- Samuel Adebayo
Centre for Intelligent Autonomous Manufacturing Systems, Queen's University Belfast, Belfast, U.K.
4 shared publications - Seán McLoone
Centre for Intelligent Autonomous Manufacturing Systems, Queen's University Belfast, Belfast, U.K.
4 shared publications Alan Cummins
Affiliation School of Psychology, Queen’s University Belfast, Belfast, Northern Ireland, United Kingdom
2 shared publications- John Douglas Crawford
Centre for Vision Research, York University, Toronto, Canada
2 shared publications Sinéad A. Reid
School of Psychology, Queen’s University Belfast, David Keir Building, 18-30 Malone Road, BT9 5BN, Belfast, Northern Ireland
2 shared publications- Theofilos Ch. Valkanidis
Affiliation School of Psychology, Queen’s University Belfast, Belfast, Northern Ireland, United Kingdom
2 shared publications
Latest Publications
- Open Access
How Robot Kinematics Influence Human Performance in Virtual Robot-to-Human Handover Tasks
Recent advancements in robotics have increased the possibilities for integrating robotic systems into human-involved workplaces, highlighting the need to examine and optimize human-robot coordination in collaborative settings. This study explores human-robot interactions during handover tasks using Virtual Reality (VR) to investigate differences in human motor performance across various task dynamics and robot kinematics. A VR-based robot handover simulation afforded safe and controlled assessments of human-robot interactions. In separate experiments, four potential influences on human performance were examined (1) control over task initiation and robot movement synchrony (temporal and spatiotemporal); (2) partner appearance (human versus robotic); (3) robot velocity profiles (minimum jerk, constant velocity, constant acceleration, and biphasic); and (4) the timing of rotational object motion. Findings across experiments emphasize humans benefit from robots providing early and salient visual information about task-relevant object motion, and advantages of human-like smooth robot trajectories. To varying degrees, these manipulations improved predictive accuracy and synchronization during interaction. This suggests that human-robot interactions should be designed to allow humans to leverage their natural capabilities for detecting biological motion, which conversely may reduce the need for costly robotic computations or added cognitive adaptation on the human side.
Open AccessHow Robot Kinematics Influence Human Performance in Virtual Robot-to-Human Handover Tasks
Recent advancements in robotics have increased the possibilities for integrating robotic systems into human-involved workplaces, highlighting the need to examine and optimize human-robot coordination in collaborative settings. This study explores human-robot interactions during handover tasks using Virtual Reality (VR) to investigate differences in human motor performance across various task dynamics and robot kinematics. A VR-based robot handover simulation afforded safe and controlled assessments of human-robot interactions. In separate experiments, four potential influences on human performance were examined (1) control over task initiation and robot movement synchrony (temporal and spatiotemporal); (2) partner appearance (human versus robotic); (3) robot velocity profiles (minimum jerk, constant velocity, constant acceleration, and biphasic); and (4) the timing of rotational object motion. Findings across experiments emphasize humans benefit from robots providing early and salient visual information about task-relevant object motion, and advantages of human-like smooth robot trajectories. To varying degrees, these manipulations improved predictive accuracy and synchronization during interaction. This suggests that human-robot interactions should be designed to allow humans to leverage their natural capabilities for detecting biological motion, which conversely may reduce the need for costly robotic computations or added cognitive adaptation on the human side.
- 18 November 2025
- Published by Springer Nature
- p. 1-51
Dyadic Human-Robot Interaction: Emerging Technologies, Challenges, and Opportunities
- 18 November 2025
- Published by Springer Nature
- p. 1-51
Dyadic Human-Robot Interaction: Emerging Technologies, Challenges, and Opportunities
- 2 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Human-Machine Systems
- Vol. 55 (3) , 333-346
SLYKLatent: A Learning Framework for Gaze Estimation Using Deep Facial Feature Learning
In this research, we present self-learn your key latent (SLYKLatent), a novel approach for enhancing gaze estimation by addressing appearance instability challenges in datasets due to aleatoric uncertainties, covariant shifts, and test domain generalization. SLYKLatent utilizes self-supervised learning for initial training with facial expression datasets, followed by refinement with a patch-based tribranch network and an inverse explained variance weighted training loss function. Our evaluation on benchmark datasets achieves a 10.98% improvement on Gaze360, supersedes the top result with 3.83% improvement on MPIIFaceGaze, and leads on a subset of ETH-XGaze by 11.59%, surpassing existing methods by significant margins. In addition, adaptability tests on RAF-DB and Affectnet show 86.4% and 60.9% accuracies, respectively. Ablation studies confirm the effectiveness of SLYKLatent's novel components. This approach has strong potential in human–robot interaction.
- 2 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Human-Machine Systems
- Vol. 55 (3) , 333-346
SLYKLatent: A Learning Framework for Gaze Estimation Using Deep Facial Feature Learning
In this research, we present self-learn your key latent (SLYKLatent), a novel approach for enhancing gaze estimation by addressing appearance instability challenges in datasets due to aleatoric uncertainties, covariant shifts, and test domain generalization. SLYKLatent utilizes self-supervised learning for initial training with facial expression datasets, followed by refinement with a patch-based tribranch network and an inverse explained variance weighted training loss function. Our evaluation on benchmark datasets achieves a 10.98% improvement on Gaze360, supersedes the top result with 3.83% improvement on MPIIFaceGaze, and leads on a subset of ETH-XGaze by 11.59%, surpassing existing methods by significant margins. In addition, adaptability tests on RAF-DB and Affectnet show 86.4% and 60.9% accuracies, respectively. Ablation studies confirm the effectiveness of SLYKLatent's novel components. This approach has strong potential in human–robot interaction.
- Open Access
- 2 December 2024
- Published by Center for Open Science in PsyArXiv
Predictive vs. Prospective Control: Clarifying the Terms of the Debate
A central question in motor control is whether that control is predictive or prospective. However, these terms are used widely and inconsistently throughout the literature, which makes it hard for research to reliably build or be challenged. Our goal is to clearly identify the ways in which these terms get used, and to identify what each approach commits to as a theory of motor control. We have identified three notions of predictive control (Perception-Action, Predictive Processing, and the Free Energy Principle) and two of prospective control (Perceptual Control Theory, and Ecological Psychology), and here we briefly discuss what each entails, with reference to key literature and examples. We hope that this guide will help researchers be clear about exactly which approach they are taking, and how they have engaged with the key elements of their chosen approach.
- 2 December 2024
- Published by Center for Open Science in PsyArXiv
Open AccessPredictive vs. Prospective Control: Clarifying the Terms of the Debate
A central question in motor control is whether that control is predictive or prospective. However, these terms are used widely and inconsistently throughout the literature, which makes it hard for research to reliably build or be challenged. Our goal is to clearly identify the ways in which these terms get used, and to identify what each approach commits to as a theory of motor control. We have identified three notions of predictive control (Perception-Action, Predictive Processing, and the Free Energy Principle) and two of prospective control (Perceptual Control Theory, and Ecological Psychology), and here we briefly discuss what each entails, with reference to key literature and examples. We hope that this guide will help researchers be clear about exactly which approach they are taking, and how they have engaged with the key elements of their chosen approach.
- Open Access
- 23 October 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 12, 157050-157066
QUB-PHEO: A Visual-Based Dyadic Multi-View Dataset for Intention Inference in Collaborative Assembly
QUB-PHEO introduces a visual-based, dyadic dataset with the potential of advancing human-robot interaction (HRI) research in assembly operations and intention inference. This dataset captures rich multimodal interactions between two participants, one acting as a ’robot surrogate,’ across a variety of assembly tasks that are further broken down into 36 distinct subtasks. With rich visual annotations—facial landmarks, gaze, hand movements, object localization, and more—for 70 participants, QUB-PHEO offers two versions: full video data for 50 participants and visual cues for all 70. Designed to improve machine learning models for HRI, QUB-PHEO enables deeper analysis of subtle interaction cues and intentions, promising contributions to the field. The dataset is available at https://github.com/exponentialR/QUB-PHEO subject to an End-User License Agreement (EULA).
- 23 October 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 12, 157050-157066
Open AccessQUB-PHEO: A Visual-Based Dyadic Multi-View Dataset for Intention Inference in Collaborative Assembly
QUB-PHEO introduces a visual-based, dyadic dataset with the potential of advancing human-robot interaction (HRI) research in assembly operations and intention inference. This dataset captures rich multimodal interactions between two participants, one acting as a ’robot surrogate,’ across a variety of assembly tasks that are further broken down into 36 distinct subtasks. With rich visual annotations—facial landmarks, gaze, hand movements, object localization, and more—for 70 participants, QUB-PHEO offers two versions: full video data for 50 participants and visual cues for all 70. Designed to improve machine learning models for HRI, QUB-PHEO enables deeper analysis of subtle interaction cues and intentions, promising contributions to the field. The dataset is available at https://github.com/exponentialR/QUB-PHEO subject to an End-User License Agreement (EULA).
- Open Access
QUB-PHEO: A Visual-Based Dyadic Multi-View Dataset for Intention Inference in Collaborative Assembly
QUB-PHEO introduces a visual-based, dyadic dataset with the potential of advancing human-robot interaction (HRI) research in assembly operations and intention inference. This dataset captures rich multimodal interactions between two participants, one acting as a 'robot surrogate,' across a variety of assembly tasks that are further broken down into 36 distinct subtasks. With rich visual annotations, such as facial landmarks, gaze, hand movements, object localization, and more for 70 participants, QUB-PHEO offers two versions: full video data for 50 participants and visual cues for all 70. Designed to improve machine learning models for HRI, QUB-PHEO enables deeper analysis of subtle interaction cues and intentions, promising contributions to the field. The dataset will be available at https://github.com/exponentialR/QUB-PHEO subject to an End-User License Agreement (EULA).
Open AccessQUB-PHEO: A Visual-Based Dyadic Multi-View Dataset for Intention Inference in Collaborative Assembly
QUB-PHEO introduces a visual-based, dyadic dataset with the potential of advancing human-robot interaction (HRI) research in assembly operations and intention inference. This dataset captures rich multimodal interactions between two participants, one acting as a 'robot surrogate,' across a variety of assembly tasks that are further broken down into 36 distinct subtasks. With rich visual annotations, such as facial landmarks, gaze, hand movements, object localization, and more for 70 participants, QUB-PHEO offers two versions: full video data for 50 participants and visual cues for all 70. Designed to improve machine learning models for HRI, QUB-PHEO enables deeper analysis of subtle interaction cues and intentions, promising contributions to the field. The dataset will be available at https://github.com/exponentialR/QUB-PHEO subject to an End-User License Agreement (EULA).
- Open Access
- 31 December 2020
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 15 (12) , e0243287
A goalkeeper’s performance in stopping free kicks reduces when the defensive wall blocks their initial view of the ball
Free kicks are an important goal scoring opportunity in football. It is an unwritten rule that the goalkeeper places a wall of defending players with the aim of making scoring harder for the attacking team. However, the defensive wall can occlude the movements of the kicker, as well as the initial part of the ball trajectory. Research on one-handed catching suggests that a ball coming into view later will likely delay movement initiation and possibly affect performance. Here, we used virtual reality to investigate the effect of the visual occlusion of the initial ball trajectory by the wall on the performance of naïve participants and skilled goalkeepers. We showed that movements were initiated significantly later when the wall was present, but not by the same amount as the duration of occlusion (~200ms, versus a movement delay of ~70-90ms); movements were thus initiated sooner after the ball came into view, based on less accumulated information. For both naïve participants and skilled goalkeepers this delayed initiation significantly affected performance (i.e., 3.6cm and 1.5cm larger spatial hand error, respectively, not differing significantly between the groups). These performance reductions were significantly larger for shorter flight times, reaching increased spatial errors of 4.5cm and 2.8cm for both groups, respectively. Further analyses showed that the wall-induced performance reduction did not differ significantly between free kicks with and without sideward curve. The wall influenced early movement biases, but only for free kicks with curve in the same direction as the required movement; these biases were away from the final ball position, thus hampering performance. Our results cannot suggest an all-out removal of the wall–this study only considered one potential downside–but should motivate goalkeepers to continuously evaluate whether placing a wall is their best option. This seems most pertinent when facing expert free kick takers for whom the wall does not act as a block (i.e., whose kicks consistently scale the wall).
- 31 December 2020
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 15 (12) , e0243287
Open AccessA goalkeeper’s performance in stopping free kicks reduces when the defensive wall blocks their initial view of the ball
Free kicks are an important goal scoring opportunity in football. It is an unwritten rule that the goalkeeper places a wall of defending players with the aim of making scoring harder for the attacking team. However, the defensive wall can occlude the movements of the kicker, as well as the initial part of the ball trajectory. Research on one-handed catching suggests that a ball coming into view later will likely delay movement initiation and possibly affect performance. Here, we used virtual reality to investigate the effect of the visual occlusion of the initial ball trajectory by the wall on the performance of naïve participants and skilled goalkeepers. We showed that movements were initiated significantly later when the wall was present, but not by the same amount as the duration of occlusion (~200ms, versus a movement delay of ~70-90ms); movements were thus initiated sooner after the ball came into view, based on less accumulated information. For both naïve participants and skilled goalkeepers this delayed initiation significantly affected performance (i.e., 3.6cm and 1.5cm larger spatial hand error, respectively, not differing significantly between the groups). These performance reductions were significantly larger for shorter flight times, reaching increased spatial errors of 4.5cm and 2.8cm for both groups, respectively. Further analyses showed that the wall-induced performance reduction did not differ significantly between free kicks with and without sideward curve. The wall influenced early movement biases, but only for free kicks with curve in the same direction as the required movement; these biases were away from the final ball position, thus hampering performance. Our results cannot suggest an all-out removal of the wall–this study only considered one potential downside–but should motivate goalkeepers to continuously evaluate whether placing a wall is their best option. This seems most pertinent when facing expert free kick takers for whom the wall does not act as a block (i.e., whose kicks consistently scale the wall).
- Open Access
- 18 December 2019
- Published by Center for Open Science in SportRxiv
A goalkeeper’s performance in stopping free kicks reduces when the defensive wall blocks their initial view of the ball
Free kicks are an important goal scoring opportunity in football. It is an unwritten rule that the goalkeeper places a wall of defending players with the aim of making scoring harder for the attacking team. However, the defensive wall can occlude the movements of the kicker, as well as the initial part of the ball trajectory. Research on one-handed catching suggests that a ball coming into view later will likely delay movement initiation and possibly affect performance. Here, we used virtual reality to investigate the effect of the visual occlusion of the initial ball trajectory by the wall on the performance of naïve participants and skilled goalkeepers. We showed that movements were initiated significantly later when the wall was present, but not by the same amount as the duration of occlusion (~200ms, versus a movement delay of ~70-90ms); movements were thus initiated sooner after the ball came into view, based on less accumulated information. For both naïve participants and skilled goalkeepers this delayed initiation significantly affected performance (i.e., 3.6cm and 1.5cm larger spatial hand error, respectively, not differing significantly between the groups). These performance reductions were significantly larger for shorter flight times, reaching increased spatial errors of 4.5cm and 2.8cm for both groups, respectively. Further analyses showed that the wall-induced performance reduction did not differ significantly between free kicks with and without sideward curve. The wall induced early movement biases, but only for free kicks with curve in the same direction as the required movement; these biases were away from the final ball position, thus hampering performance. Our results cannot suggest an all-out removal of the wall - this study only considered one potential downside – but should motivate goalkeepers to continuously evaluate whether placing a wall is their best option. This seems most pertinent when facing expert free kick takers for whom the wall does not act as a block (i.e., whose kicks consistently scale the wall).
- 18 December 2019
- Published by Center for Open Science in SportRxiv
Open AccessA goalkeeper’s performance in stopping free kicks reduces when the defensive wall blocks their initial view of the ball
Free kicks are an important goal scoring opportunity in football. It is an unwritten rule that the goalkeeper places a wall of defending players with the aim of making scoring harder for the attacking team. However, the defensive wall can occlude the movements of the kicker, as well as the initial part of the ball trajectory. Research on one-handed catching suggests that a ball coming into view later will likely delay movement initiation and possibly affect performance. Here, we used virtual reality to investigate the effect of the visual occlusion of the initial ball trajectory by the wall on the performance of naïve participants and skilled goalkeepers. We showed that movements were initiated significantly later when the wall was present, but not by the same amount as the duration of occlusion (~200ms, versus a movement delay of ~70-90ms); movements were thus initiated sooner after the ball came into view, based on less accumulated information. For both naïve participants and skilled goalkeepers this delayed initiation significantly affected performance (i.e., 3.6cm and 1.5cm larger spatial hand error, respectively, not differing significantly between the groups). These performance reductions were significantly larger for shorter flight times, reaching increased spatial errors of 4.5cm and 2.8cm for both groups, respectively. Further analyses showed that the wall-induced performance reduction did not differ significantly between free kicks with and without sideward curve. The wall induced early movement biases, but only for free kicks with curve in the same direction as the required movement; these biases were away from the final ball position, thus hampering performance. Our results cannot suggest an all-out removal of the wall - this study only considered one potential downside – but should motivate goalkeepers to continuously evaluate whether placing a wall is their best option. This seems most pertinent when facing expert free kick takers for whom the wall does not act as a block (i.e., whose kicks consistently scale the wall).
- Open Access
- 17 May 2018
- Published by Springer Nature in Scientific Reports
- Vol. 8 (1) , 7756
Non-predictive online spatial coding in the posterior parietal cortex when aiming ahead for catching
Catching movements must be aimed ahead of the moving ball, which may require predictions of when and where to catch. Here, using repetitive Transcranial Magnetic Stimulation we show for the first time that the Superior Parietal Occipital Cortex (SPOC) displays non-predictive online spatial coding at the moment the interception movements were already aimed at the predicted final target position. The ability to aim ahead for catching must thus arise downstream within the parietofrontal network for reaching.
- 17 May 2018
- Published by Springer Nature in Scientific Reports
- Vol. 8 (1) , 7756
Open AccessNon-predictive online spatial coding in the posterior parietal cortex when aiming ahead for catching
Catching movements must be aimed ahead of the moving ball, which may require predictions of when and where to catch. Here, using repetitive Transcranial Magnetic Stimulation we show for the first time that the Superior Parietal Occipital Cortex (SPOC) displays non-predictive online spatial coding at the moment the interception movements were already aimed at the predicted final target position. The ability to aim ahead for catching must thus arise downstream within the parietofrontal network for reaching.
- 1 January 2018
- Published by American Psychological Association (APA) in Journal of Experimental Psychology: Human Perception and Performance
- Vol. 44 (1) , 38-52
Spatial biases in motion extrapolation for manual interception.
The exact mechanisms by which humans control the manual interception of moving targets are currently unknown. Here, the authors explored the behaviors associated with the spatial control for manual interception. The examined task required controlling a cursor to intercept moving targets on a touch screen. They explored the effects of target motion direction, curvature and occlusion on manual interception. They observed occlusion-dependent spatial errors and arrival times for curved and diagonal trajectories (larger errors and earlier arrival of the finger at its final position with longer occlusion). These effects were particularly apparent for targets moving away from screen center at interception due to curve. In a follow-up experiment, the authors showed that the outward curve effects on spatial errors were absent because the associated trajectories appears to move toward positions that participants could expect the target to never reach. Their analyses also revealed occlusion-dependent spatial errors for diagonal trajectories, which is the well-known angle-of-approach effect. Follow-up experiments demonstrated that this effect was not due to the central initial cursor position acting as a visual reference point or the initial ocular pursuit. Most importantly, the angle-of-approach effect persisted in a judgment task. The authors thus concluded that this effect does not stem from online information-based modulations of movement speed, but from target information used to control aiming (i.e., movement direction). Moreover, processing for diagonal target motion appears to be biased toward straight downward. Public Significance Statement This study examines the control of manual interception, for a range of target trajectories, using visual occlusion. The authors show that occlusion causes spatial biases in the movements because unseen target motion is not fully accounted for. Participants quite accurately intercepted targets moving on straight trajectories and targets continuously visible; spatial biases arose, however, when unseen target motion must be accounted for. Because these effects were present irrespective of the time pressure inherent to manual interception, the authors interpret these to originate from target information used to control where to move rather than how fast to move. This research has implications in sports training, suggesting that the usefulness of visual occlusion training may be dependent on exactly how occluded objects are moving.
- 1 January 2018
- Published by American Psychological Association (APA) in Journal of Experimental Psychology: Human Perception and Performance
- Vol. 44 (1) , 38-52
Spatial biases in motion extrapolation for manual interception.
The exact mechanisms by which humans control the manual interception of moving targets are currently unknown. Here, the authors explored the behaviors associated with the spatial control for manual interception. The examined task required controlling a cursor to intercept moving targets on a touch screen. They explored the effects of target motion direction, curvature and occlusion on manual interception. They observed occlusion-dependent spatial errors and arrival times for curved and diagonal trajectories (larger errors and earlier arrival of the finger at its final position with longer occlusion). These effects were particularly apparent for targets moving away from screen center at interception due to curve. In a follow-up experiment, the authors showed that the outward curve effects on spatial errors were absent because the associated trajectories appears to move toward positions that participants could expect the target to never reach. Their analyses also revealed occlusion-dependent spatial errors for diagonal trajectories, which is the well-known angle-of-approach effect. Follow-up experiments demonstrated that this effect was not due to the central initial cursor position acting as a visual reference point or the initial ocular pursuit. Most importantly, the angle-of-approach effect persisted in a judgment task. The authors thus concluded that this effect does not stem from online information-based modulations of movement speed, but from target information used to control aiming (i.e., movement direction). Moreover, processing for diagonal target motion appears to be biased toward straight downward. Public Significance Statement This study examines the control of manual interception, for a range of target trajectories, using visual occlusion. The authors show that occlusion causes spatial biases in the movements because unseen target motion is not fully accounted for. Participants quite accurately intercepted targets moving on straight trajectories and targets continuously visible; spatial biases arose, however, when unseen target motion must be accounted for. Because these effects were present irrespective of the time pressure inherent to manual interception, the authors interpret these to originate from target information used to control where to move rather than how fast to move. This research has implications in sports training, suggesting that the usefulness of visual occlusion training may be dependent on exactly how occluded objects are moving.