INVITED SPEAKERS
The following speakers tentatively confirmed their participation in the workshop.
Modeling and synthesizing expressive movements in HRI and pHRI
Abstract: TBD
Active inference in human-robot interaction
Abstract: When humans collaborate on a common task while being mechanically connected, such as carrying a table together, they modulate their muscle activation to extract maximum information from the partner. My presentation will present this novel finding, and describe i) how this can be modelled using nonlinear stochastic optimal control, and ii) how this can give rise to contact robots optimally adapting their control to their human user.
Design and analysis of external devices with musculo-skeletal models
Abstract: With their ability to estimate internal and interface loads, musculoskeletal models provide an excellent tool to study the interaction between the musculoskeletal system and an external device. This can be beneficial for the analysis of external devices, such as an exoskeleton or orthosis, or in multiple parts of the design phase of external devices. In this part of the workshop, we will look into how we can use the models to get insights into the fundamental mechanism that we can use for idea generation of the external device, virtual prototyping, adjustment of the device to subject-specific characteristics, and evaluation of internal body loads when the device has been developed. The workshop will showcase the development of a novel knee orthosis and demonstrate the AnyBody Modeling System used throughout the simulations.
Modeling and simulating the human-exoskeleton interaction
Abstract: TBD
Embodied Sensorimotor Intelligence
Abstract: Embodied Movements are a readout of "Embodied Sensorimotor Intelligence". A mantra that underlines most scientific efforts and advancements in our field can be summarized as -- Gradual improvements in the understanding of movements will incrementally unlock the fundamentals of Embodied Sensorimotor Intelligence.
Such an incremental approach has been challenged in multiple fields such as vision, natural language, games, protein, etc, leading to transformative advancements unlocked by a direct pursuit of the problem in its full complexity. Sidestepping motion studies, this talk will delve into strategies for a direct pursuit of Embodied Motor Intelligence. What will such a pursuit entail? What technological/ practical barriers are on the way? And what opportunities and new capabilities can be unlocked? What standing problems can be solved?
Optimal linear quadratic regulator control for assisting human balance with physiological time delays
Abstract: Falling is the leading cause of injury in elderly individuals. Exoskeletons can compensate for muscle weakness and reflex delays, helping to maintain balance by applying assistive forces to the body. The dynamics of systems containing time delays are governed by delay differential equations (DDEs) which present analytical challenges, particularly in the context of optimal control. In this work, we convert DDEs into systems of ordinary differential equations using the Galerkin projection technique. We then use the so-called “Averaging” method to determine the optimal gains for a linear quadratic regulator (LQR) controller that applies assistive torques at the ankle joint of an inverted-pendulum human model. We verify our formulation with two examples from the literature, then apply this strategy to study exoskeleton-assisted human balance where physiological torques from the plantar flexor and dorsiflexor muscles are generated based on delayed state feedback. We show that the exoskeleton torques provided by the LQR controller prevent the loss of balance of inherently unstable human models. Finally, we use the utopian point method to demonstrate the practical application of our computational framework in analyzing the tradeoff between the system’s settling time and the power consumed by the exoskeleton.
All Human Models are Wrong but Some are Useful: Synergizing Learning- and Model-Based Control for Advancing Human-Robot Interaction
Abstract: TBD
Reciprocal Dynamics in Physical Human-Robot Interaction: Influences on Human Motor Control and Feedback for Robot Control
Abstract: In this talk, I will explore how both humans and robots can be made more adaptable through enhanced sensory feedback and reward mechanisms. I will discuss computational models that are crucial for understanding of sensorimotor and reward learning and the related neural correlates that can be used to develop responsive robot behaviors leading to effective physical human-robot interaction.
On human-in-the-loop optimization of human-robot interaction
Abstract: From industrial exoskeletons to implantable medical devices, robots that interact closely with people are poised to improve every aspect of our lives. Yet designing these systems is very challenging; humans are incredibly complex, and in many cases we respond to robotic devices in ways that cannot be sufficiently modeled or predicted. Human-in-the-loop optimization can overcome these challenges by systematically, empirically identifying the device characteristics that result in the best objective performance for a specific user and application. This approach has enabled dramatic improvements in human-robot performance in research settings and has the potential to speed development and enhance products. In this talk, I will discuss opportunities for developing new human-in-the-loop optimization techniques, with an emphasis on the potential role of data-driven models. My aim is to inspire workshop participants to develop effective new tools and use them to design new robotic devices that truly enhance the human experience.