AgiBot GO-1: The Evolution of Generalist Embodied Foundation Model from VLA to ViLLA

Quick Summary:

VLM: Using the InternVL-2B model, it processes multi-view images, force signals and language inputs to provide scene understanding and instruction comprehension.2. Latent Planner: This expert predicts Latent Action Tokens based on intermediate outputs from the VLM, forming a Chain of Planning (CoP) for general action understanding and planning.3. Action Expert: It generates the final fine-grained action sequences based on intermediate outputs from the VLM and the Latent Action Tokens.  The following is an introduction to the two key components of MoE: Latent Planner and Action Expert.  Expert 1: Latent Planner Although the AgiBot World dataset is the largest real world robot dataset globally, the volume of action-labeled robot data remains limited relative to internet-scale datasets.

The MoE consists of two key components: the Latent Planner, which learns from cross-embodiment and human operation data to develop general action understanding, and the Action Expert, which uses over a million real robot demonstrations to achieve high-frequency and dexterous manipulation.  These components work in synergy, providing GO-1's unique capabilities： Learning from Human VideosFew-shot GeneralizationCross-Embodiment AdaptationContinuous Self-Evolution Paper:  YouTube Link: At the end of 2024, AgiBot launched the AgiBot World dataset, a large-scale, high-quality real world robotics dataset comprising over 1 million trajectories across 217 tasks in five application domains.

Compared to the Vision-Language-Action (VLA) model, where actions are directly conditioned on vision and language inputs, the ViLLA model predicts latent action tokens, bridging the gap between image-text inputs and robot actions generated by the action expert.

Read Quick SummarySave Time

or

Article Details

Author / Journalist: AgiBot

Category: Technology

Markets: China

Topics: World Expert Go-1 Data Understanding Latent Robot Action Framework Agibot Model

Source Website Secure: Yes (HTTPS)

News Sentiment: Neutral

Fact Checked: Legitimate

Article Type: News Report

Published On: 2025-03-11 @ 12:45:00 (1 days ago)

News Timezone: GMT +8:00

News Source URL: globenewswire.com

Language: English

Article Length: 639 words

Reading Time: 4 minutes read

Sentences: 30 lines

Sentence Length: 22 words per sentence (average)

Platforms: Desktop Web, Mobile Web, iOS App, Android App

Copyright Owner: © GlobeNewswire

News ID: 26891803

View Article Analysis

About GlobeNewswire

Main Topics: Technology

Official Website: globenewswire.com

Update Frequency: 62 posts per day

Year Established: 1998

Headquarters: China

News Last Updated: 8 hours ago

Coverage Areas: China

Ownership: Independent Company

Publication Timezone: GMT +8:00

Content Availability: Worldwide

News Language: English

RSS Feed: Available (XML)

API Access: Available (JSON, REST)

Website Security: Secure (HTTPS)

Publisher ID: #116

Publisher Details

Frequently Asked Questions