Force Generative Imitation Learning: Bridging Position Trajectory and Force Commands through Control Technique
Hiroshi Sato, Sho Sakaino, Toshiaki Tsuji
TL;DR
This work tackles the challenge of generating force commands for contact-rich manipulation by introducing force generative imitation learning that converts position trajectories into force commands. It deploys a hierarchical architecture with a memory-based upper layer and a memoryless lower layer, combined with a PID feedback loop to ensure stable tracking. The approach demonstrates improved accuracy and generalization on a character-writing task using a CRANE-X7 manipulator, showing that force-aware control can be achieved without memory-induced instability. Time-scale separation between the upper and lower layers enables robust, memoryless control while leveraging learned trajectory predictions for force generation.
Abstract
In contact-rich tasks, while position trajectories are often easy to obtain, appropriate force commands are typically unknown. Although it is conceivable to generate force commands using a pretrained foundation model such as Vision-Language-Action (VLA) models, force control is highly dependent on the specific hardware of the robot, which makes the application of such models challenging. To bridge this gap, we propose a force generative model that estimates force commands from given position trajectories. However, when dealing with unseen position trajectories, the model struggles to generate accurate force commands. To address this, we introduce a feedback control mechanism. Our experiments reveal that feedback control does not converge when the force generative model has memory. We therefore adopt a model without memory, enabling stable feedback control. This approach allows the system to generate force commands effectively, even for unseen position trajectories, improving generalization for real-world robot writing tasks.