Carlo D’Eramo

Adaptive Reward Design in Reinforcement Learning: A Taxonomy and Survey

Raphaela Baybas, Carlo D’Eramo, Philipp Brune

Adaptive Reward Design (ARD) is becoming a fundamental component for Reinforcement Learning (RL) agents, as they are deployed in increasingly complex settings where a single static reward across all phases of learning is rarely sufficient. Yet, ARD is rarely studied as a …

de (code pays fourni par la source)

Rationality Randomization With Maximum Entropy for Robust Dynamic Obstacle Avoidance on Legged Robots

Jose-Luis Holgado-Alvarez, Gabriele Tiboni, Aryaman Reddi, Carlo D’Eramo

de (code pays fourni par la source)

Self-Paced Curriculum Reinforcement Learning for Autonomous Superbike Racing in Simulation

Luca Ghisi, Jacopo Essenziale, Carlo D’Eramo, Matteo Luperto

Autonomous Racing has seen remarkable progress through deep Reinforcement Learning (RL), primarily for four-wheeled vehicles. However, motorbikes introduce substantially greater complexity due to the need to manage balance and lean angle, in addition to more reactive steering and throttle control, and a …

Self-Paced Curriculum Reinforcement Learning for Autonomous Superbike Racing in Simulation

Luca Ghisi, Jacopo Essenziale, Carlo D’Eramo, Matteo Luperto

Autonomous Racing has seen remarkable progress through deep Reinforcement Learning (RL), primarily for four-wheeled vehicles. However, motorbikes introduce substantially greater complexity due to the need to manage balance and lean angle, in addition to more reactive steering and throttle control, and a …

it, de (code pays fourni par la source)

HARBOR: A Harness Framework for Agentic Robot Reinforcement Learning

Zechu Li, Yufeng Jin, Xiaoyang Liu, Puze Liu et autres

Reinforcement learning (RL) has become a powerful paradigm for robot learning, particularly in sim-to-real settings, but its broader adoption remains limited by the engineering pipeline surrounding the algorithms. Building tasks, shaping rewards, and tuning hyperparameters require substantial expert effort, making RL workflows …

HARBOR: A Harness Framework for Agentic Robot Reinforcement Learning

Zechu Li, Yufeng Jin, Xiaoyang Liu, Puze Liu et autres

Reinforcement learning (RL) has become a powerful paradigm for robot learning, particularly in sim-to-real settings, but its broader adoption remains limited by the engineering pipeline surrounding the algorithms. Building tasks, shaping rewards, and tuning hyperparameters require substantial expert effort, making RL workflows …

Streaming Reinforcement Learning under Partial Observability with Real-Time Recurrent Learning

Noah Farr, Aryaman Reddi, Carlo D’Eramo, Jan Peters

Streaming reinforcement learning has emerged as an online learning paradigm that conforms to the restrictions of natural learning agents that process data incrementally, i.e. with a batch size of 1 and no replay buffer. While streaming RL has recently been shown to …

Streaming Reinforcement Learning under Partial Observability with Real-Time Recurrent Learning

Noah Farr, Aryaman Reddi, Carlo D’Eramo, Jan Peters

Streaming reinforcement learning has emerged as an online learning paradigm that conforms to the restrictions of natural learning agents that process data incrementally, i.e. with a batch size of 1 and no replay buffer. While streaming RL has recently been shown to …

de (code pays fourni par la source)

Do Not Imitate, Reinforce: Iterative Classification via Belief Refinement

Mahdi Kallel, Johannes Tölle, Ahmed Hendawy, Carlo D’Eramo

Standard supervised classification trains models to imitate the exact labels provided by a perfect oracle. This imitation happens in a single pass, restricting the model to a fixed compute budget even when inputs vary in complexity. Moreover, the rigid training objective forces …

Do Not Imitate, Reinforce: Iterative Classification via Belief Refinement

Mahdi Kallel, Johannes Tölle, Ahmed Hendawy, Carlo D’Eramo

Standard supervised classification trains models to imitate the exact labels provided by a perfect oracle. This imitation happens in a single pass, restricting the model to a fixed compute budget even when inputs vary in complexity. Moreover, the rigid training objective forces …

de, ca (code pays fourni par la source)

Continual Model-based Reinforcement Learning for Fault-Adaptive Control

Arash Torabi Goodarzi, Waxenegger-Wilfing Günther, Carlo D’Eramo

Use the Online Network If You Can: Towards Fast and Stable Reinforcement Learning

Ahmed Hendawy, Henrik Metternich, Théo Vincent, Jan Peters et autres

The use of target networks is a popular approach for estimating value functions in deep Reinforcement Learning (RL). While effective, the target network remains a compromise solution that preserves stability at the cost of slowly moving targets, thus delaying learning. Conversely, using …