Small decisions. Interesting possibilities.Submit contentSubmit

A dual-arm robot with Jev in the middle layer

型付き意思決定AIモデル「Jev」で双腕ロボを動かした結果まとめてみた🦾 3層制御のうち意思決定(2層)をJevに任せ、IKや物理計算はコード側に分離する構造を作りました。500msの応答と0.5円/試行で早くて激安。

Added to Jevfast

How it uses Jev

The post says Jev handles the decision layer of a three-layer robot controller, while inverse kinematics and physics calculations remain in code.

App workflow

  1. Read the robot task state

    Read the robot task state.

  2. Ask Jev for a decision in the middle control layer

    Ask Jev for a decision in the middle control layer.

  3. Use code for inverse kinematics and physics

    Use code for inverse kinematics and physics.

  4. Apply the resulting commands to the dual-arm robot

    Apply the resulting commands to the dual-arm robot.

Why it is interesting

The design assigns semantic decisions to Jev while retaining precise motion and physics calculations in conventional code.