Loris Roveda

Angelo Moroncelli, Matteo Rufolo, Gunes Cagin Aydin et al.

Accurate modeling of robot dynamics is essential for model-based control, yet remains challenging under distributional shifts and real-time constraints. In this work, we formulate system identification as an in-context meta-learning problem and compare deterministic and generative sequence models for forward dynamics prediction. We take a Transformer-based meta-model, as a strong deterministic baseline, and introduce to this setting two complementary diffusion-based approaches: (i) inpainting diffusion (Diffuser), which learns the joint input-observation distribution, and (ii) conditioned diffusion models (CNN and Transformer), which generate future observations conditioned on control inputs. Through large-scale randomized simulations, we analyze performance across in-distribution and out-of-distribution regimes, as well as computational trade-offs relevant for control. We show that diffusion models significantly improve robustness under distribution shift, with inpainting diffusion achieving the best performance in our experiments. Finally, we demonstrate that warm-started sampling enables diffusion models to operate within real-time constraints, making them viable for control applications. These results highlight generative meta-models as a promising direction for robust system identification in robotics.

Angelo Moroncelli, Roberto Zanetti, Marco Maccarini et al.

Reinforcement learning (RL) enables high-frequency, closed-loop control for robotic manipulation, but scaling to long-horizon tasks with sparse or imperfect rewards remains difficult due to inefficient exploration and poor credit assignment. Vision-Language-Action (VLA) models leverage large-scale multimodal pretraining to provide generalist, task-level reasoning, but current limitations hinder their direct use in fast and precise manipulation. In this paper, we propose Vision-Language-Action Jump-Starting (VLAJS), a method that bridges sparse VLA guidance with on-policy RL to improve exploration and learning efficiency. VLAJS treats VLAs as transient sources of high-level action suggestions that bias early exploration and improve credit assignment, while preserving the high-frequency, state-based control of RL. Our approach augments Proximal Policy Optimization (PPO) with a directional action-consistency regularization that softly aligns the RL agent's actions with VLA guidance during early training, without enforcing strict imitation, requiring demonstrations, or relying on continuous teacher queries. VLA guidance is applied sparsely and annealed over time, allowing the agent to adapt online and ultimately surpass the guiding policy. We evaluate VLAJS on six challenging manipulation tasks: lifting, pick-and-place, peg reorientation, peg insertion, poking, and pushing in simulation, and validate a subset on a real Franka Panda robot. VLAJS consistently outperforms PPO and distillation-style baselines in sample efficiency, reducing required environment interactions by over 50% in several tasks. Real-world experiments demonstrate zero-shot sim-to-real transfer and robust execution under clutter, object variation, and external perturbations.

Le Anh Dao, Loris Roveda, Marco Maccarini et al.

Black-box optimization refers to the optimization problem whose objective function and/or constraint sets are either unknown, inaccessible, or non-existent. In many applications, especially with the involvement of humans, the only way to access the optimization problem is through performing physical experiments with the available outcomes being the preference of one candidate with respect to one or many others. Accordingly, the algorithm so-called Active Preference Learning has been developed to exploit this specific information in constructing a surrogate function based on standard radial basis functions, and then forming an easy-to-solve acquisition function which repetitively suggests new decision vectors to search for the optimal solution. Based on this idea, our approach aims to extend the algorithm in such a way that can exploit further information effectively, which can be obtained in reality such as: 5-point Likert type scale for the outcomes of the preference query (i.e., the preference can be described in not only "this is better than that" but also "this is much better than that" level), or multiple outcomes for a single preference query with possible additive information on how certain the outcomes are. The validation of the proposed algorithm is done through some standard benchmark functions, showing a promising improvement with respect to the state-of-the-art algorithm.

Angelo Moroncelli, Vishal Soni, Marco Forgione et al.

Recently, generative AI and reinforcement learning (RL) have been redefining what is possible for AI agents that take information flows as input and produce intelligent behavior. As a result, we are seeing similar advancements in embodied AI and robotics for control policy generation. Our review paper examines the integration of generative AI models with RL to advance robotics. Our primary focus is on the duality between generative AI and RL for robotics downstream tasks. Specifically, we investigate: (1) The role of prominent generative AI tools as modular priors for multi-modal input fusion in RL tasks. (2) How RL can train, fine-tune and distill generative models for policy generation, such as VLA models, similarly to RL applications in large language models. We then propose a new taxonomy based on a considerable amount of selected papers. Lastly, we identify open challenges accounting for model scalability, adaptation and grounding, giving recommendations and insights on future research directions. We reflect on which generative AI models best fit the RL tasks and why. On the other side, we reflect on important issues inherent to RL-enhanced generative policies, such as safety concerns and failure modes, and what are the limitations of current methods. A curated collection of relevant research papers is maintained on our GitHub repository, serving as a resource for ongoing research and development in this field: https://github.com/clmoro/Robotics-RL-FMs-Integration.

Jannick van Buuren, Roberto Giglio, Loris Roveda et al.

This paper explores how deliberate mutations of reward function in reinforcement learning can produce diversified skill variations in robotic manipulation tasks, examined with a liquid pouring use case. To this end, we developed a new reward function mutation framework that is based on applying Gaussian noise to the weights of the different terms in the reward function. Inspired by the cost-benefit tradeoff model from human motor control, we designed the reward function with the following key terms: accuracy, time, and effort. The study was performed in a simulation environment created in NVIDIA Isaac Sim, and the setup included Franka Emika Panda robotic arm holding a glass with a liquid that needed to be poured into a container. The reinforcement learning algorithm was based on Proximal Policy Optimization. We systematically explored how different configurations of mutated weights in the rewards function would affect the learned policy. The resulting policies exhibit a wide range of behaviours: from variations in execution of the originally intended pouring task to novel skills useful for unexpected tasks, such as container rim cleaning, liquid mixing, and watering. This approach offers promising directions for robotic systems to perform diversified learning of specific tasks, while also potentially deriving meaningful skills for future tasks.