Skip to main content

# Long-Horizon Robot Manipulation

2 related articles

Subscribe to updates

arXiv Oct 5, 2026
PAM encodes long-horizon intent as proprioceptive sketches, raising success from 47.5% to 75.0% on four real-world bimanual tasks
The work proposes Proprioceptive Action Models (PAM), which jointly generate a compact, timing-free sketch of the robot's remaining joint-space path and a dense executable action chunk within a single transformer denoiser, with the sketch parameterized by arc length rather than time to capture geometric intent invariant to execution timing and with block-causal attention and a staggered denoising schedule maintaining directed sketch-to-action dependence; in simulation PAM improves over its action-only counterparts on Push-T and LIBERO-Long, and on four real-world bimanual tasks it raises success from 47.5% to 75.0%.
The work proposes Proprioceptive Action Models (PAM), which jointly generate a compact, timing-free sketch of the robot's remaining joint-space path and a dense executable action chunk within a single transformer denoiser, with the sketch parameterized by arc length rather than time to capture geometric intent invariant to execution timing and with block-causal attention and a staggered denoising schedule maintaining directed sketch-to-action dependence; in simulation PAM improves over its action-only counterparts on Push-T and LIBERO-Long, and on four real-world bimanual tasks it raises success from 47.5% to 75.0%.
Fangyuan Wang（First author）
Proprioceptive Action Models Proprioceptive Sketch Long-Horizon Robot Manipulation
arXiv Oct 1, 2026
RoboCoach turns imagined failures into targeted demos: 150 extra subtask demonstrations lift Franka success from 13.3% to 75.0%
The work presents RoboCoach, a world-model-guided active coaching framework whose Route-Imagine-Diagnose-Improve (RIDI) loop executes reusable skill experts in closed loop inside CoachWorld, a shared action-conditioned world model, uses a progress judge to record the first subtask that fails to complete, and thereby selects which subtask demonstrations to acquire and which expert adapters to update; across two simulation suites and two real-robot platforms, imagined and deployed success correlate with Spearman rho = 0.840 over 22 task-policy pairs, and with only 150 additional subtask demonstrations per platform success rises from 13.3% to 75.0% on Franka and from 40.0% to 83.8% on AgileX, while the coached experts reach 35.
The work presents RoboCoach, a world-model-guided active coaching framework whose Route-Imagine-Diagnose-Improve (RIDI) loop executes reusable skill experts in closed loop inside CoachWorld, a shared action-conditioned world model, uses a progress judge to record the first subtask that fails to complete, and thereby selects which subtask demonstrations to acquire and which expert adapters to update; across two simulation suites and two real-robot platforms, imagined and deployed success correlate with Spearman rho = 0.840 over 22 task-policy pairs, and with only 150 additional subtask demonstrations per platform success rises from 13.3% to 75.0% on Franka and from 40.0% to 83.8% on AgileX, while the coached experts reach 35.
Beijing Academy of Artificial Intelligence University of Nottingham Tsinghua University Renmin University of China
Jiajun Liu（First author）
RoboCoach CoachWorld World Model Long-Horizon Robot Manipulation

You’re all caught up