Article

Open-source vision-language-action navigation model LightNav-0 integrated into Microduck MuJoCo simulation: instruction in, walking out on Jetson edge devices

The author integrated the open-source vision-language navigation model LightNav-0 into the Microduck simulation (MuJoCo). The robot receives instructions such as "walk to the wooden door frame and stop," predicts a path from first-person frames, and executes it via MPC and the official ONNX gait policy.

Part of the event: LightNav-0 Integrated into MuJoCo Simulation Demo for Closed-Loop VLN Navigation

This page provides a summary and source link only.