Hybrid control improves robot book insertion in tight shelves
Hybrid Residual Reinforcement Learning for Contact-Rich Robotic Book Insertion
Robotics
Summary
Putting a book into a tight shelf is tricky because tiny errors can cause the book to get stuck or not sit right. The authors studied how a robot can use a mix of fixed rules about the shelf’s shape and a learning system that adjusts its movements locally. Their new method keeps a main controller for general movements but adds a learning part that makes small, careful corrections and decides when to let go of the book. Tests showed this approach worked much better than using fixed rules alone, both in simulations and real robots.
What this means in practice
- •For robotics engineers: Develop better robot controllers for placing objects precisely in tight spaces with improved success and reduced failures.
- •For manufacturing automation teams: Enhance automated assembly lines that require inserting parts into close-fitting slots to reduce jamming and increase throughput.
Authors
Tianyuan Liu, Rutherford Agbeshi Patamia, Benjamin Champion, Akansel Cosgun, Richard Dazeley
Abstract
Placing a grasped book into a tight shelf is a compact but difficult contact-rich control problem: millimetre-scale pose error can turn a geometrically valid approach into jamming, failed release, or incomplete seating. We study this final phase after grasp acquisition and global approach, and ask how control authority should be divided between known geometry and learned behaviour. Our method retains a nominal task-space controller for structured insertion and seating, while residual PPO supplies bounded local corrections and decides when to release. Only the brief open-retreat-reclose transition is scripted. For the final policy used on hardware, a deployment-matched simulation evaluation over 512 fixed conditions yields 98.50 percent mean success (0.23 percentage-point sample SD) across three independent training runs, compared with 37.89 percent for nominal control. On the physical xArm7, 60 trials over 30 matched conditions show the same qualitative advantage: residual control raises success from 26.7 percent to 63.3 percent, reduces failures from 22 to 11, and wins 13 of the 15 matched conditions in which the two controllers differ. Robustness tests show that performance remains above 87 percent under initialization perturbations up to 1.5x, while very tight clearances expose the geometric limit of local correction. These results support a hybrid design in which geometry preserves reliable task structure and learning is concentrated on the contact-sensitive behaviour that fixed rules handle poorly.