AI & ML
Improving Imitation Policies for Multi-Phase Tasks through Residual Learning Allocation
arXiv cs.RO0 views
arXiv:2611.25320v1 Announce Type: cross Abstract: Residual reinforcement learning corrects an imitation-trained base policy using a limited budget of environment interaction. In a task with several phases, this budget must be divided among the phases, and the phase in which the base policy fails is not necessarily the phase in which a correction most improves task completion. We formalize this distinction through the completion gain: the change in full-task completion probability when, during a single phase, a state-feedback controller replaces selected components of the base policy's action. We decompose the completion gain into the probability of reaching the phase and the conditional improvement in completion from the start of the phase, and we estimate it from paired episodes. The positive estimates determine which action components the residual policy corrects and how training episodes are distributed across phases. In stochastic point-mass docking experiments, most failures occurred in the final docking phase, whereas the largest completion gain occurred in the preceding channel phase; gain-based allocation improved completion over allocation proportional to failure counts and, in a variant with a longer approach, over uniform allocation at half of the budget. In simulated bimanual robotic assembly, the completion-gain estimates were positive only for insertion and release, although few failures occurred during release; in a second assembly configuration, a residual policy trained in the release phase improved completion from held-out states at the start of release by 16.9%.
Source:https://arxiv.org/abs/2611.25320PreviousCrisis Governance at the Joints. Mapping the Ethical and Societal Territory of Crisis Management Technologies: The Case of Human-Swarm TeamingNextFlowDrive: moderated flow matching with data balancing for trajectory planning
Related posts

On modularization
0dev.to
Is there a standard for AI agent receipts? Yes. Five systems bent around it in one day.
0dev.to
Installed but never connects: five platforms, five different permission prompts
0dev.to
465 practice tests, 423 pages, and every single test without a page of its own is a full sitting
0dev.to
Sort a Month of Stripe Failures by Code Before You Change Anything
0dev.to
How we find viral moments in a 6-hour stream in 4 minutes
0dev.to