EN 
Українська
PolskiVision-Language-Action (VLA) Robot Simulator (2D)
Top-down 2D VLA robot demo: a vision encoder locates objects, a language encoder parses a typed command, and an action decoder solves a 2-link inverse-kinematics trajectory the arm then executes.
This 2D companion drives the same vision → language → action pipeline as the 3D version through a top-down canvas view: a command is parsed into a target object and a destination box, a 2-link inverse-kinematics solver plans the arm's shoulder and elbow angles, and the arm executes a pick-and-place trajectory while a live readout shows both joint angles in degrees.
Top-down 2D VLA robot demo: vision encoder locates objects, language encoder parses a typed command, and a 2-link inverse-kinematics action decoder plans and executes the pick-and-place trajectory.
2D · HTML5 Canvas 2D · 60 FPS target · runs fully client-side, no install