Create

Sign in to ReadmeX

or

How we handle your data: Privacy Policy

ReadmeX

Your AI reading companion

What would you like to catch up on?

Pick a briefing, or ask your own question.

Your community and people briefings stay personal to you.

Story

Ant Lingbo open-sources LingBot-Touch tactile foundation model

AI summary

Ant Lingbo Technology open-sourced LingBot-Touch, a tactile foundation model that encodes magnetic, visuotactile and piezoresistive sensor signals into a shared tactile representation for robotic manipulation. The company said the model was pretrained on nearly 5,000 hours of real tactile data. In its own evaluation across 26 real-robot tasks, four action models reportedly achieved average success rates about 14 to 26 percentage points higher with the model than without tactile information.

In-depth insight

The model is positioned as a general tactile interface between different sensors, dexterous hands and action models, allowing tactile data and features to be reused across hardware and policy combinations and reducing repeated adaptation work.

11月10日,蚂蚁灵波科技开源触觉基础模型 LingBot-Touch,为机器人大脑补齐关键的触觉感知能力。该模型基于近5000小时真实触觉数据预训练,覆盖磁触觉、视触觉和压阻式三类主流传感原理,可将不同信号转化为同一套触觉表征,供不同灵巧手和动作模型使用,让机器人在“看得清楚、想得明白”的同时,进一步补齐精细操作所需的触觉感知。

Hugging Face 地址:https://huggingface.co/collections/robbyant/lingbot-touch

GitHub 地址:https://github.com/Robbyant/lingbot-touch

技术报告:https://technology.robbyant.com/lingbot-touch

在通用大模型加速演进的时点,选择推出一款面向物理操作闭环的触觉基础模型,传递出蚂蚁灵波对具身原生路线的持续投入和笃定。

#Ant Lingbo Technology#LingBot-Touch#LingBot-Touch-Tiny#LingBot-Touch-Small#LingBot-VLA#π0.5#Diffusion Policy#T-Rex#Sharpa Wave#XHand1

Original · 6 min readRead the original
Group chat

No comments yet. Start the conversation.