TacForcing: Streaming Action Generation with Execution-Time Tactile Feedback
技术动态
来源：arXiv cs.RO发布时间待核实
arXiv:2608.25798v1 Announce Type: new Abstract: Contact-rich manipulation requires adapting to contact states that can evolve substantially within an action horizon. However, chunk-based vision-language-action models predict complete action chunks from observations collected before execution, leaving tactile conditioning stale during execution. Existing tactile-reactive approaches typically rely on separate high-frequency controllers, which increase both architectural and training complexity.