LEARNING DEVICE AND ROBOT CONTROL SYSTEM
To provide a learning device which can make a robot perform some operation even when an operation instruction newly input to the robot is not clear.SOLUTION: A learning device 40 comprises: a learning unit 42 which learns measurement information, a photographed image and instruction information by u...
Gespeichert in:
Hauptverfasser: | , , |
---|---|
Format: | Patent |
Sprache: | eng ; jpn |
Schlagworte: | |
Online-Zugang: | Volltext bestellen |
Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
Zusammenfassung: | To provide a learning device which can make a robot perform some operation even when an operation instruction newly input to the robot is not clear.SOLUTION: A learning device 40 comprises: a learning unit 42 which learns measurement information, a photographed image and instruction information by using a model read out from a model definition unit 41 according to a purpose operation; a weight storage unit 43 which stores the weight to the model learned by the learning unit 42; and an inference unit 44 which applies the weight read out from the weight storage unit 43 to the model read out from the model definition unit 41 when the instruction information 30 is newly input to the robot 10, predicts an intention of the instruction information by applying the measurement information that is newly measured, the photographed image that is newly captured and the instruction information 30 that is newly input to the robot 10 to the model, infers the purpose operation of the robot 10 and an object being a target of the purchase operation, and predicts a drive command for making the robot 10 perform the purpose operation according to the predicted intention of the instruction information.SELECTED DRAWING: Figure 2
【課題】ロボットに対して新たに入力された動作指示が明確でなくてもロボットに何らかの動作を行わせることが可能な学習装置を提供する。【解決手段】学習装置40は、目的動作に合わせてモデル定義部41から読み出したモデルを用いて、計測情報と、撮影画像と、指示情報とを学習する学習部42と、学習部42が学習したモデルへの重みを保存する重み保存部43と、ロボット10への指示情報30が新たに入力されると、モデル定義部41から読み出したモデルに重み保存部43から読み出した重みをかけ、新たに計測された計測情報と、新たに撮影された撮影画像と、新たに入力されたロボット10へ指示情報30とをモデルに適用して指示情報の意図を予測し、かつロボット10の目的動作、及び目的動作の対象となる物体を推論し、予測した指示情報の意図に合わせた目的動作をロボット10に行わせるための駆動指令を予測する推論部44を備える。【選択図】図2 |
---|