Whole-Body Semantic-to-Actuation Grounding of Elephant-Inspired Soft-Trunk Motion via Lightweight Flow Matching

2026-07-12

AI Summary

This work addresses the challenge of efficiently mapping open-vocabulary semantic commands to whole-body motions of embodied soft robotic arms, where end-effector–centric control lacks expressiveness and high-dimensional instructions often violate physical feasibility. To bridge this gap, the authors propose a lightweight flow-matching framework that leverages a multimodal large language model to generate morphology-aware intent–intensity tuples, parameterizes tendon trajectories via Catmull-Rom splines, and employs a rectified flow generator to sample physically plausible whole-body motions. This approach achieves the first end-to-end, lightweight semantic-to-actuation mapping, boosting command grounding accuracy from 25.0% to 77.2% while reducing inference time to 4.87 ms. In human–robot interaction trials involving 100 participants, user satisfaction significantly increased from 46% to 82%, demonstrating both technical efficacy and practical usability.

Abstract

For close-contact human-robot interaction (HRI), trunk-like continuum manipulators provide a physical channel for diverse whole-body expression, but grounding open-vocabulary responses into such robots is difficult: end-effector motion underspecifies body shape, whereas direct whole-body commands are high-dimensional and hard to keep feasible. We propose a whole-body semantic-to-actuation grounding framework for elephant-inspired soft-trunk HRI based on lightweight flow matching. The framework converts responses from a multimodal large language model into bounded, morphology-aligned intent-intensity tuples, parameterizes tendon-actuation trajectories with compact Catmull-Rom spline controls, and uses a rectified-flow generator to sample feasible whole-body trunk motions. Experiments show that the proposed framework improves held-out grounding correctness from 25.0% to 77.2% over a raw-response dense-regression baseline. Compared with a denoising-diffusion baseline, it improves correctness from 71.9% to 77.2% and reduces inference time from 7.86 ms to 4.87 ms while preserving motion diversity. A 100-participant physical HRI study further shows that adding the generated soft-trunk motion channel increases the positive overall-satisfaction rating from 46% to 82% over the audiovisual-only baseline.

Problem

• semantic-to-actuation grounding
• soft-trunk robot
• whole-body motion
• human-robot interaction
• continuum manipulator

Innovation

• flow matching
• semantic-to-actuation grounding
• soft trunk robot
• whole-body motion generation
• lightweight motion synthesis