GitLab Transcend - Intelligent orchestration
Mini toolsVision Force Action (VFA)
By Eka Robotics
Eka’s Vision-Force-Action, or VFA, model is designed to close the gap between robotic generality and real-world dexterous performance. According to the launch coverage, it uses vision plus force-aware control as the basis for manipulation, with the company arguing that force is the “native language” of the physical world. Eka says the model is trained through large-scale, high-fidelity simulation rather than human imitation data, with the goal of transferring skills from sim to real for fast, delicate tasks such as assembly, improvised sorting, and tactile grip correction.
New
Image
Gen 4
Overview
Eka’s Vision-Force-Action model is a robotics foundation model built for high-speed, contact-rich dexterity. Instead of relying mainly on language or imitation, it treats force as the core signal for physical interaction, enabling delicate and fast manipulation tasks such as screwing in light bulbs, handling soft food, and recovering from slips in real time.
About Eka Robotics
Eka Robotics is an AI-native robotics startup developing the Vision-Force-Action (VFA) foundation model — a new class of model that treats force as the native language of the physical world and trains in high-fidelity simulation rather than from human imitation video.
Industry:
Artificial Intelligence
Company Size:
25
Location:
Cambridge, Massachusetts, US
Tools using Vision Force Action (VFA)
No tools found for this model yet.
Last updated: May 3, 2026
