Schedule + Papers

Welcome and Opening Remarks for R4P

Invited Talk I: Yiannis Demiris

“Assistive Robots: from robot learning to human empowerment.”

Assistive robots hold great potential for empowering people to achieve their intended tasks, for example during activities of daily living. However, every person is unique, creating challenges for the robots’ perceptual, cognitive and motor systems, and necessitating the development of interactive learning algorithms that personalise the robot’s assistance to each individual. In this talk, I will outline research in our laboratory towards the development of algorithms that enable learning during human-robot interaction, both for robots as well as for humans; I will demonstrate their application in activities of daily living, and illustrate how computer vision, learning and augmented reality can enable adaptive user interfaces for interacting with assistive robots in a safe and trustworthy manner.

Invited Talk II: Stefanie Tellex

“Towards complex language in partially observed environments.”

Robots can act as a force multiplier for people, whether a robot assisting an astronaut with a repair on the International Space station, a UAV taking flight over our cities, or an autonomous vehicle driving through our streets. Existing approaches use action-based representations that do not capture the goal-based meaning of a language expression and do not generalize to partially observed environments. The aim of my research program is to create autonomous robots that can understand complex goal-based commands and execute those commands in partially observed, dynamic environments. I will describe demonstrations of object-search in a POMDP setting with information about object locations provided by language, and mapping between English and Linear Temporal Logic, enabling a robot to understand complex natural language commands in city-scale environments. These advances represent steps towards robots that interpret complex natural language commands in partially observed environments using a decision theoretic framework.

Break

Poster session (gather.town)

1. Game-theoretic Model of Trust to Infer Human’s Observation Strategy of Robot Behavior

2. Leveraging Semantic Scene Graphs and Explainable AI to Explain Robot Failures

3. Human Action Nodes for Behavior Trees

4. Sampling-Based Robust Control of Autonomous Systems with Non-Gaussian Noise

5. Dialogue Object Search

6. Leveraging Temporal Structure in Task Specifications for POMDP Planning

7. Safe Human-Interactive Control via Shielding

8. Simultaneously Learning Human Preferences and Environment Dynamics Parameters

9. Towards Interactively Improving Human Users' Understanding of Robot Behavior

10. Constrained Feedforward Neural Network Training via Reachability Analysis

11. Reactive Synthesis for Human-aware Robotic Manipulation using Regret Games

12. DULA: A Differentiable Ergonomics Model for Postural Optimization in pHRI

13. Information-Theoretic Based Target Search with Multiple Agents

14. Robust Policies for Uncertain POMDPs

15. Towards Explainable Multi-robot Motion Planning

16. Safe Control with Neural Network Dynamic Models

Interactive Lunch session (gather.town)

Invited Talk III: Changliu Liu

“Safety-critical learning and control for collaborative robots.”

This talk will share some of our recent work that enables autonomous robotic systems to safely operate in uncertain and human-involved environments. The safety specification can be written as constraints on the system's state space. To ensure that these constraints are satisfied throughout time, the robot needs to correctly anticipate the future and only select the action that will not lead to a state that violates the constraints. To deal with the uncertainties, the robot needs to continuously learn the environment dynamics and adjust its behavior accordingly. This solution strategy requires seamless integration between set-theoretic control and continual learning. This talk will focus on two aspects of the problem: 1) how to perform provably safe control in real time with learned models and 2) how to achieve data-efficient learning. For the first aspect, I will introduce a safe control method that ensures forward invariance inside the safety constraint with black-box dynamic models (e.g., deep neural networks). For the second aspect, I will introduce a verification-guided learning method that performs more learning on most vulnerable parts of the model. The computations that involve deep neural networks are handled by our toolbox NeuralVerification.jl, a sound verification toolbox that can check input-output properties of deep neural networks. I will conclude the talk with future visions.

Invited Talk IV: Dan Bohus

“Situated Interaction.”

Situated language interaction is a complex, multimodal affair that extends well beyond the spoken word. When interacting, we use a wide array of non-verbal signals and incrementally coordinate with each other to simultaneously resolve several problems: we manage engagement, coordinate on taking turns, recognize intentions, and establish and maintain common ground as a basis for contributing to the conversation. Proximity and body pose, attention and gaze, head nods and hand gestures, prosody and facial expressions, all play very important roles in this process. And just like a couple of decades ago advances in speech recognition opened up the field of spoken dialog systems, current advances in vision and other perceptual technologies are again opening up new horizons -- we are starting to be able to build machines that computationally understand these social signals and the physical world around them, and participate in physically situated interactions and collaborations with people. In this talk, using a number of research vignettes from work we have done over the last decade at Microsoft Research, I will draw attention to some of the challenges and opportunities that lie ahead of us in this exciting space. In particular, I will discuss issues with managing engagement and turn-taking in multiparty open-world settings, and more generally highlight the importance of timing and fine-grained coordination in situated language interaction. Finally, I will conclude by describing an open-source framework we are developing that promises to simplify the construction of physically situated interactive systems, and in the process further enable and accelerate research in this area.

Break

Closing Remarks