Dimensions overall score 8.0

GitHub Code Pulse

No public code linked for this paper yet.

• FG-CLTP is a fine-grained contrastive language tactile pretraining framework that incorporates quantitative contact states.method 95%
• A novel dataset with over 100k tactile 3D point cloud-language pairs capturing multidimensional contact states was created.method 95%
• A discretized numerical tokenization mechanism is used for quantitative-semantic alignment in FG-CLTP.method 95%
• FG-CLTP achieves 95.9% classification accuracy.result 95%
• FG-CLTP reduces regression error (MAE) by 52.6% compared to state-of-the-art methods.result 95%
• The integration of 3D point cloud representations in FG-CLTP results in a minimal sim-to-real gap of 3.5%.technical 95%
• A 3D tactile-language-action (3D-TLA) architecture driven by a flow matching policy is developed based on FG-CLTP.method 95%
• The proposed framework significantly outperforms strong baselines in contact-rich manipulation tasks.result 90%

Competitive landscape

Competitor map is still being generated for this paper. Enable generation or check back soon.

Startup potential card