analiza aiqbench
This paper studies reinforcement learning for robotic arm reaching, showing that adding an analytic potential-field term to the reward can give policies a safety margin around obstacles they cannot observe. The results come from simulation with a Franka Panda, and the margin is a fixed buffer rather than something that adapts to obstacle size, which limits how far the findings extend. For ML readers, the useful point is that reward design can encode safety constraints without expanding the policy's inputs, though the comparison to a classical planner that knows the obstacle positions needs careful reading.
Positive Safety Margins from Artificial Potential Field Reward Augmentation for Size-Robust Obstacle Avoidance under Partial Observability
arXiv:2611.02947v1 Announce Type: new Abstract: Collision-free reaching is central to manipulation, yet real cells often leave the size and exact position of an obstacle unobserved, which excludes reinforcement learning controllers that take obstacle geometry as input. A reward limited to a collision penalty does not depend on the clearance outside the obstacle, so under a step cost its learned policy skims the obstacle surface and fails once the obstacle radius exceeds its training value. We investigate whether a single analytic reward term can induce a safety margin without exposing the obstacle to the policy. We augment the reward of a simulated Franka Emika Panda reaching task with a horizontal artificial potential field that penalizes proximity within a safety margin around each of two pillars (fixed nominal layout, random drift). Under hidden observation, policies trained with this term maintain a safe success rate above 88% across a radius sweep from 0.03 m to 0.055 m (30 seeds), while policies without it fall below 10% at the maximum tested radius; the term induces a minimum clearance of 47.9 mm against 9.6 mm without it and eliminates training collisions at least eight times earlier (median steps). The advantage is statistically significant at all out-of-distribution radii and holds across three reinforcement learning solvers and under full observation. Its safe success rate is comparable to that of a classical potential-field planner given the pillar positions. The margin is a fixed buffer rather than an adaptation to the radius, and it lengthens the path by about 19%.
Abstract:Collision-free reaching is central to manipulation, yet real cells often leave the size and exact position of an obstacle unobserved, which excludes reinforcement learning controllers that take obstacle geometry as input. A reward limited to a collision penalty does not depend on the clearance outside the obstacle, so under a step cost its learned policy skims the obstacle surface and fails once the obstacle radius exceeds its training value. We investigate whether a single analytic reward term can induce a safety margin without exposing the obstacle to the policy. We augment the reward of a simulated Franka Emika Panda reaching task with a horizontal artificial potential field that penalizes proximity within a safety margin around each of two pillars (fixed nominal layout, random drift). Under hidden observation, policies trained with this term maintain a safe success rate above 88% across a radius sweep from 0.03 m to 0.055 m (30 seeds), while policies without it fall below 10% at the maximum tested radius; the term induces a minimum clearance of 47.9 mm against 9.6 mm without it and eliminates training collisions at least eight times earlier (median steps). The advantage is statistically significant at all out-of-distribution radii and holds across three reinforcement learning solvers and under full observation. Its safe success rate is comparable to that of a classical potential-field planner given the pillar positions. The margin is a fixed buffer rather than an adaptation to the radius, and it lengthens the path by about 19%.
44 pages, 15 figures. Code and data: this https URL
Robotics (cs.RO)
arXiv:2611.02947 [cs.RO]
(or arXiv:2611.02947v1 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2611.02947 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 349 cuvinte
Subiecte conexe
Harnessing VLM in Embodied RL: From Reward Regressor to Success Detector via Exploration-Augmented Key-Observation Gating
arXiv · 14 oct. 2026
Evaluating Machine Learning Models for Post-Wildfire Debris-Flow Prediction
arXiv · 7 aug. 2026
A Novel Approach to Evaluating Evaluation Metrics for Multi-Output Structured Prediction
arXiv · 30 oct. 2026
Policy-Conditioned Counterfactual Credit for Verifiable Reinforcement Learning of Long-Horizon Language Agents
arXiv · 6 iun. 2026