Mental Modeling of Reinforcement Learning Agents by Language Models

Can emergent language models faithfully model the intelligence of decision-making agents? Though modern language models exhibit already some reasoning ability, and theoretically can potentially express any probable distribution over tokens, it remains underexplored how the world knowledge these pretrained models have memorized can be utilized to comprehend an agent's behaviour in the physical world. This study empirically examines, for the first time, how well large language models (LLMs) can build a mental model of agents, termed agent mental modelling, by reasoning about an agent's behaviour and its effect on states from agent interaction history. This research may unveil the potential of leveraging LLMs for elucidating RL agent behaviour, addressing a key challenge in eXplainable reinforcement learning (XRL). To this end, we propose specific evaluation metrics and test them on selected RL task datasets of varying complexity, reporting findings on agent mental model establishment. Our results disclose that LLMs are not yet capable of fully mental modelling agents through inference alone without further innovations. This work thus provides new insights into the capabilities and limitations of modern LLMs.

Details Make a Difference: Object State-Sensitive Neurorobotic Task Planning

The state of an object reflects its current status or condition and is important for a robot's task planning and manipulation. However, detecting an object's state and generating a state-sensitive plan for robots is challenging. Recently, pre-trained Large Language Models (LLMs) and Vision-Language Models (VLMs) have shown impressive capabilities in generating plans. However, to the best of our knowledge, there is hardly any investigation on whether LLMs or VLMs can also generate object state-sensitive plans. To study this, we introduce an Object State-Sensitive Agent (OSSA), a task-planning agent empowered by pre-trained neural networks. We propose two methods for OSSA: (i) a modular model consisting of a pre-trained vision processing module (dense captioning model, DCM) and a natural language processing model (LLM), and (ii) a monolithic model consisting only of a VLM. To quantitatively evaluate the performances of the two methods, we use tabletop scenarios where the task is to clear the table. We contribute a multimodal benchmark dataset that takes object states into consideration. Our results show that both methods can be used for object state-sensitive tasks, but the monolithic approach outperforms the modular approach. The code for OSSA is available at \url{https://github.com/Xiao-wen-Sun/OSSA}

Agentic Skill Discovery

Language-conditioned robotic skills make it possible to apply the high-level reasoning of Large Language Models (LLMs) to low-level robotic control. A remaining challenge is to acquire a diverse set of fundamental skills. Existing approaches either manually decompose a complex task into atomic robotic actions in a top-down fashion, or bootstrap as many combinations as possible in a bottom-up fashion to cover a wider range of task possibilities. These decompositions or combinations, however, require an initial skill library. For example, a "grasping" capability can never emerge from a skill library containing only diverse "pushing" skills. Existing skill discovery techniques with reinforcement learning acquire skills by an exhaustive exploration but often yield non-meaningful behaviors. In this study, we introduce a novel framework for skill discovery that is entirely driven by LLMs. The framework begins with an LLM generating task proposals based on the provided scene description and the robot's configurations, aiming to incrementally acquire new skills upon task completion. For each proposed task, a series of reinforcement learning processes are initiated, utilizing reward and success determination functions sampled by the LLM to develop the corresponding policy. The reliability and trustworthiness of learned behaviors are further ensured by an independent vision-language model. We show that starting with zero skill, the ASD skill library emerges and expands to more and more meaningful and reliable skills, enabling the robot to efficiently further propose and complete advanced tasks. The project page can be found at: https://agentic-skill-discovery.github.io.

Unified Dynamic Scanpath Predictors Outperform Individually Trained Neural Models

Previous research on scanpath prediction has mainly focused on group models, disregarding the fact that the scanpaths and attentional behaviors of individuals are diverse. The disregard of these differences is especially detrimental to social human-robot interaction, whereby robots commonly emulate human gaze based on heuristics or predefined patterns. However, human gaze patterns are heterogeneous and varying behaviors can significantly affect the outcomes of such human-robot interactions. To fill this gap, we developed a deep learning-based social cue integration model for saliency prediction to instead predict scanpaths in videos. Our model learned scanpaths by recursively integrating fixation history and social cues through a gating mechanism and sequential attention. We evaluated our approach on gaze datasets of dynamic social scenes, observed under the free-viewing condition. The introduction of fixation history into our models makes it possible to train a single unified model rather than the resource-intensive approach of training individual models for each set of scanpaths. We observed that the late neural integration approach surpasses early fusion when training models on a large dataset, in comparison to a smaller dataset with a similar distribution. Results also indicate that a single unified model, trained on all the observers' scanpaths, performs on par or better than individually trained models. We hypothesize that this outcome is a result of the group saliency representations instilling universal attention in the model, while the supervisory signal and fixation history guide it to learn personalized attentional behaviors, providing the unified model a benefit over individual models due to its implicit representation of universal attention.

Comparing Apples to Oranges: LLM-powered Multimodal Intention Prediction in an Object Categorization Task

Intention-based Human-Robot Interaction (HRI) systems allow robots to perceive and interpret user actions to proactively interact with humans and adapt to their behavior. Therefore, intention prediction is pivotal in creating a natural interactive collaboration between humans and robots. In this paper, we examine the use of Large Language Models (LLMs) for inferring human intention during a collaborative object categorization task with a physical robot. We introduce a hierarchical approach for interpreting user non-verbal cues, like hand gestures, body poses, and facial expressions and combining them with environment states and user verbal cues captured using an existing Automatic Speech Recognition (ASR) system. Our evaluation demonstrates the potential of LLMs to interpret non-verbal cues and to combine them with their context-understanding capabilities and real-world knowledge to support intention prediction during human-robot interaction.

Inverse Kinematics for Neuro-Robotic Grasping with Humanoid Embodied Agents

This paper introduces a novel zero-shot motion planning method that allows users to quickly design smooth robot motions in Cartesian space. A B\'ezier curve-based Cartesian plan is transformed into a joint space trajectory by our neuro-inspired inverse kinematics (IK) method CycleIK, for which we enable platform independence by scaling it to arbitrary robot designs. The motion planner is evaluated on the physical hardware of the two humanoid robots NICO and NICOL in a human-in-the-loop grasping scenario. Our method is deployed with an embodied agent that is a large language model (LLM) at its core. We generalize the embodied agent, that was introduced for NICOL, to also be embodied by NICO. The agent can execute a discrete set of physical actions and allows the user to verbally instruct various different robots. We contribute a grasping primitive to its action space that allows for precise manipulation of household objects. The new CycleIK method is compared to popular numerical IK solvers and state-of-the-art neural IK methods in simulation and is shown to be competitive with or outperform all evaluated methods when the algorithm runtime is very short. The grasping primitive is evaluated on both NICOL and NICO robots with a reported grasp success of 72% to 82% for each robot, respectively.

Diffusing in Someone Else's Shoes: Robotic Perspective Taking with Diffusion

Humanoid robots can benefit from their similarity to the human shape by learning from humans. When humans teach other humans how to perform actions, they often demonstrate the actions and the learning human can try to imitate the demonstration. Being able to mentally transfer from a demonstration seen from a third-person perspective to how it should look from a first-person perspective is fundamental for this ability in humans. As this is a challenging task, it is often simplified for robots by creating a demonstration in the first-person perspective. Creating these demonstrations requires more effort but allows for an easier imitation. We introduce a novel diffusion model aimed at enabling the robot to directly learn from the third-person demonstrations. Our model is capable of learning and generating the first-person perspective from the third-person perspective by translating the size and rotations of objects and the environment between two perspectives. This allows us to utilise the benefits of easy-to-produce third-person demonstrations and easy-to-imitate first-person demonstrations. The model can either represent the first-person perspective in an RGB image or calculate the joint values. Our approach significantly outperforms other image-to-image models in this task.

Large Language Models for Orchestrating Bimanual Robots

Although there has been rapid progress in endowing robots with the ability to solve complex manipulation tasks, generating control policies for bimanual robots to solve tasks involving two hands is still challenging because of the difficulties in effective temporal and spatial coordination. With emergent abilities in terms of step-by-step reasoning and in-context learning, Large Language Models (LLMs) have taken control of a variety of robotic tasks. However, the nature of language communication via a single sequence of discrete symbols makes LLM-based coordination in continuous space a particular challenge for bimanual tasks. To tackle this challenge for the first time by an LLM, we present LAnguage-model-based Bimanual ORchestration (LABOR), an agent utilizing an LLM to analyze task configurations and devise coordination control policies for addressing long-horizon bimanual tasks. In the simulated environment, the LABOR agent is evaluated through several everyday tasks on the NICOL humanoid robot. Reported success rates indicate that overall coordination efficiency is close to optimal performance, while the analysis of failure causes, classified into spatial and temporal coordination and skill selection, shows that these vary over tasks. The project website can be found at http://labor-agent.github.io

Human Impression of Humanoid Robots Mirroring Social Cues

Mirroring non-verbal social cues such as affect or movement can enhance human-human and human-robot interactions in the real world. The robotic platforms and control methods also impact people's perception of human-robot interaction. However, limited studies have compared robot imitation across different platforms and control methods. Our research addresses this gap by conducting two experiments comparing people's perception of affective mirroring between the iCub and Pepper robots and movement mirroring between vision-based iCub control and Inertial Measurement Unit (IMU)-based iCub control. We discovered that the iCub robot was perceived as more humanlike than the Pepper robot when mirroring affect. A vision-based controlled iCub outperformed the IMU-based controlled one in the movement mirroring task. Our findings suggest that different robotic platforms impact people's perception of robots' mirroring during HRI. The control method also contributes to the robot's mirroring performance. Our work sheds light on the design and application of different humanoid robots in the real world.

Robotic Imitation of Human Actions

Imitation can allow us to quickly gain an understanding of a new task. Through a demonstration, we can gain direct knowledge about which actions need to be performed and which goals they have. In this paper, we introduce a new approach to imitation learning that tackles the challenges of a robot imitating a human, such as the change in perspective and body schema. Our approach can use a single human demonstration to abstract information about the demonstrated task, and use that information to generalise and replicate it. We facilitate this ability by a new integration of two state-of-the-art methods: a diffusion action segmentation model to abstract temporal information from the demonstration and an open vocabulary object detector for spatial information. Furthermore, we refine the abstracted information and use symbolic reasoning to create an action plan utilising inverse kinematics, to allow the robot to imitate the demonstrated action.