Most recent successes in robot reinforcement learning involve learning a specialized single-task agent. However, robots capable of performing multiple tasks can be much more valuable in real-world applications. Multi-task reinforcement learning can be very challenging due to the increased sample complexity and the potentially conflicting task objectives. Previous work on this topic is dominated by model-free approaches. The latter can be very sample inefficient even when learning specialized single-task agents. In this work, we focus on model-based multi-task reinforcement learning. We propose a method for learning multi-task visual world models, leveraging pre-trained language models to extract semantically meaningful task representations. These representations are used by the world model and policy to reason about task similarity in dynamics and behavior. Our results highlight the benefits of using language-driven task representations for world models and a clear advantage of model-based multi-task learning over the more common model-free paradigm.

This study presents an innovative test rig engineered to explore the kinematic and viscoelastic characteristics of human specimen hands. The rig features eight force-controlled motors linked to muscle tendons, enabling precise stimulation of hand specimens. Hand movements are monitored through an optical tracking system, while a force-torque sensor quantifies the resultant fingertip loads. Employing this setup, we successfully demonstrated a pincer grip using a cadaver hand and measured both muscle forces and grip strength. Our results reveal a nonlinear relationship between tendon forces and grip strength, which can be modeled by an exponential fit. This investigation serves as a nexus between biomechanical and robotics-focused research, providing critical insights for the advancement of robotic hand actuation and therapeutic interventions.

In robotics and biomechanics, accurately determining joint parameters and computing the corresponding forward and inverse kinematics are critical yet often challenging tasks, especially when dealing with highly individualized and partly unknown systems. This paper unveils a cutting-edge kinematic optimizer, underpinned by an autoencoder-based architecture, to address these challenges. Utilizing a neural network, our approach simulates inverse kinematics, converting measurement data into joint-specific parameters during encoding, enabling a stable optimization process. These parameters are subsequently processed through a predefined, differentiable forward kinematics model, resulting in a decoded representation of the original data. Beyond offering a comprehensive solution to kinematics challenges, our method also unveils previously unidentified joint parameters. Real experimental data from knee and hand joints validate the optimizer’s efficacy. Additionally, our optimizer is multifunctional: it streamlines the modeling and automation of kinematics and enables a nuanced evaluation of diverse modeling techniques. By assessing the differences in reconstruction losses, we illuminate the merits of each approach. Collectively, this preliminary study signifies advancements in kinematic optimization, with potential applications spanning both biomechanics and robotics.

In offline reinforcement learning, a policy is learned using a static dataset in the absence of costly feedback from the environment. In contrast to the online setting, only using static datasets poses additional challenges, such as policies generating out-of-distribution samples. Model-based offline reinforcement learning methods try to overcome these by learning a model of the underlying dynamics of the environment and using it to guide policy search. It is beneficial but, with limited datasets, errors in the model and the issue of value overestimation among out-of-distribution states can worsen performance. Current model-based methods apply some notion of conservatism to the Bellman update, often implemented using uncertainty estimation derived from model ensembles. In this paper, we propose Constrained Latent Action Policies (C-LAP) which learns a generative model of the joint distribution of observations and actions. We cast policy learning as a constrained objective to always stay within the support of the latent action distribution, and use the generative capabilities of the model to impose an implicit constraint on the generated actions. Thereby eliminating the need to use additional uncertainty penalties on the Bellman update and significantly decreasing the number of gradient steps required to learn a policy. We empirically evaluate C-LAP on the D4RL and V-D4RL benchmark, and show that C-LAP is competitive to state-of-the-art methods, especially outperforming on datasets with visual observations.

Unlike most reinforcement learning agents which require an unrealistic amount of environment interactions to learn a new behaviour, humans excel at learning quickly by merely observing and imitating others. This ability highly depends on the fact that humans have a model of their own embodiment that allows them to infer the most likely actions that led to the observed behaviour. In this paper, we propose Action Inference by Maximising Evidence (AIME) to replicate this behaviour using world models. AIME consists of two distinct phases. In the first phase, the agent learns a world model from its past experience to understand its own body by maximising the ELBO. While in the second phase, the agent is given some observation-only demonstrations of an expert performing a novel task and tries to imitate the expert's behaviour. AIME achieves this by defining a policy as an inference model and maximising the evidence of the demonstration under the policy and world model. Our method is "zero-shot" in the sense that it does not require further training for the world model or online interactions with the environment after given the demonstration. We empirically validate the zero-shot imitation performance of our method on the Walker and Cheetah embodiment of the DeepMind Control Suite and find it outperforms the state-of-the-art baselines. Code is available at: https://github.com/argmax-ai/aime.

For many tasks, tactile or visual feedback is helpful or even crucial. However, designing controllers that take such high-dimensional feedback into account is non-trivial. Therefore, robots should be able to learn tactile skills through trial and error by using reinforcement learning algorithms. The input domain for such tasks, however, might include strongly correlated or non-relevant dimensions, making it hard to specify a suitable metric on such domains. Auto-encoders specialize in finding compact representations, where defining such a metric is likely to be easier. Therefore, we propose a reinforcement learning algorithm that can learn non-linear policies in continuous state spaces, which leverages representations learned using auto-encoders. We first evaluate this method on a simulated toy-task with visual input. Then, we validate our approach on a real-robot tactile stabilization task.