analiza aiqbench
CaSCo introduces a motion planning framework that integrates vision-language models and physics simulators to assess semantic risk and predict cascaded collisions in dynamic environments. By moving beyond binary collision constraints, the system optimizes robot paths to minimize potential damage from indirect object displacements, marking a significant step toward safer autonomous manipulation. This research advances the intersection of AI reasoning and robotic control, directly addressing safety challenges in cluttered physical spaces.
CaSCo: Cascade-Aware Soft-Collision Motion Planning
arXiv:2609.18910v1 Announce Type: new Abstract: Conventional motion planning treats collision as a binary constraint, although contact with different objects can have drastically different consequences. A robot may safely brush against a cardboard box while even minor contact with a glass, laptop, or unstable object may be undesirable. Moreover, a direct robot--object collision can move the contacted object and trigger secondary object--object collisions, making the risk of a motion depend on the physical evolution of the scene rather than only on the robot's geometric path. We present CaSCo, a cascade-aware soft-collision motion planning framework in which a vision-language or language model assigns semantic risk to objects and a physics simulator predicts the consequences of candidate robot motions. CaSCo searches for a path that minimizes the total semantic risk of the unique objects displaced either directly by the robot or indirectly through cascaded collisions. Because collisions change the environment, we augment roadmap states with the predicted object arrangement and the set of objects whose risk has already been incurred. We develop an optimal graph-search algorithm with an admissible and consistent cascade-relaxed heuristic and caching and pruning mechanisms for efficient search. Experiments in cluttered manipulation environments evaluate semantic risk, cascade reasoning, planning efficiency, and real-robot operation.
Abstract:Conventional motion planning treats collision as a binary constraint, although contact with different objects can have drastically different consequences. A robot may safely brush against a cardboard box while even minor contact with a glass, laptop, or unstable object may be undesirable. Moreover, a direct robot--object collision can move the contacted object and trigger secondary object--object collisions, making the risk of a motion depend on the physical evolution of the scene rather than only on the robot's geometric path. We present CaSCo, a cascade-aware soft-collision motion planning framework in which a vision-language or language model assigns semantic risk to objects and a physics simulator predicts the consequences of candidate robot motions. CaSCo searches for a path that minimizes the total semantic risk of the unique objects displaced either directly by the robot or indirectly through cascaded collisions. Because collisions change the environment, we augment roadmap states with the predicted object arrangement and the set of objects whose risk has already been incurred. We develop an optimal graph-search algorithm with an admissible and consistent cascade-relaxed heuristic and caching and pruning mechanisms for efficient search. Experiments in cluttered manipulation environments evaluate semantic risk, cascade reasoning, planning efficiency, and real-robot operation.
Robotics (cs.RO)
arXiv:2609.18910 [cs.RO]
(or arXiv:2609.18910v1 [cs.RO] for this version)
https://doi.org/10.48550/arXiv.2609.18910 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 · 291 cuvinte
Subiecte conexe
Splat-CBF: Safe Next-Best-View Control in 3D Gaussian-Splat Maps
arXiv · 22 sept. 2026
Foundation-Model-Guided Topology-Aware Semantic Risk Fields for Manipulation
arXiv · 30 sept. 2026
Agri-Sim: Agricultural Simulation Platform for Embodied Intelligence Evaluation in Greenhouse Robotics
arXiv · 2 sept. 2026
Not All History Helps: Velocity-Aware Selective Memory for Long-Horizon End-to-End Autonomous Driving
arXiv · 18 aug. 2026