JP4419035B2 - Robot apparatus and control method thereof - Google Patents

Description

【００４８】
を用いて次の周期におけるその情動のパラメータ値Ｅ〔ｔ＋１〕を算出する。
【００４９】
そして感情・本能モデル部３１は、この演算結果を現在のその情動のパラメータ値Ｅ〔ｔ〕と置き換えるようにしてその情動のパラメータ値を更新する。なお各認識結果や各出力行動に対してどの情動のパラメータ値を更新するかは、予め決められており、例えば「叩かれた」といった認識結果が与えられた場合には「怒り」の情動のパラメータ値が上がると共に「喜び」の情動のパラメータ値が下がり、「撫でられた」といった認識結果が与えられた場合には「喜び」の情動のパラメータ値が上がると共に「悲しみ」及び「怒り」の各情動のパラメータ値が下がる。
【００５０】
これと同様にして、感情・本能モデル部３１は、「運動欲」、「愛情欲」、「食欲」、「好奇心」及び「睡眠欲」の互いに独立した５つの欲求について、これら欲求ごとにその欲求の強さを表すパラメータ値を保持している。そして感情・本能モデル部３１は、これら各欲求のパラメータ値を、それぞれ状態認識部３０からの認識結果や、経過時間及び行動決定部３２からの通知などに基づいて順次更新する。
【００５１】
具体的に感情・本能モデル部３１は、「運動欲」、「愛情欲」及び「好奇心」についてはペットロボット１の出力行動、経過時間及び認識結果などに基づいて所定の演算式により算出されるその欲求の変動量ΔＩ〔ｋ〕、現在のその欲求のパラメータ値をＩ〔ｋ〕、その欲求の感度を表す係数をｋ_ｉとして、所定周期で次式
【００５２】
【数２】

【００５３】
を用いて次の周期におけるその欲求のパラメータ値Ｉ〔ｋ＋１〕を算出し、この演算結果を現在のその欲求のパラメータ値Ｉ〔ｋ〕と置き換えるようにしてその欲求のパラメータ値を更新する。この場合、出力行動や認識結果等に対してどの欲求のパラメータ値を変化させるかは予め定められており、例えば行動決定部３２から何らかの行動を発現したとの通知があった場合には「運動欲」のパラメータ値が下がる。
【００５４】
また感情・本能モデル部３１は、「食欲」については、状態認識部３０を介して与えられるバッテリ残量検出信号Ｓ２Ａ（図２）に基づいて、バッテリ残量をＢ_Ｌとして所定周期で次式
【００５５】
【数３】

【００５６】
により「食欲」のパラメータ値〔ｋ＋１〕を算出し、この演算結果を現在の食欲のパラメータ値Ｉ〔ｋ〕と置き換えるようにして当該「食欲」のパラメータ値を更新する。
【００５７】
さらに感情・本能モデル部４１は、「睡眠欲」については１日を１同期として所定時間ごとにパラメータ値を増減変更する。
【００５８】
なお本実施の形態においては、各情動及び各欲求のパラメータ値がそれぞれ０から１００までの範囲で変動するように規制されており、また係数ｋ_ｅ、ｋ_ｉの値も各情動及び各欲求ごとに個別に設定されている。
【００５９】
（２−３）行動決定部３２の構成
行動決定部３２は、メモリ１０Ａ内に複数の行動モデルを有している。そして行動決定部３２は、状態認識部３０から与えられる状態認識情報Ｓ１０と、感情・本能モデル部３１における各情動及び各欲求のパラメータ値と、対応する行動モデルと、経過時間となどに基づいて次の行動を決定し、決定結果を行動決定情報Ｓ１１として感情・本能モデル部３１、行動生成部３３及び成長制御部３４に出力する。なお行動決定部３２についての詳細は後述するので、ここでは簡単に行動モデルを利用した次の行動の決定手法について説明する。
【００６０】
このペットロボット１では、次の行動を決定する手法として、図５に示すように１つのノード（状態）ＮＯＤＥ_０から同じ又は他のどのノードＮＯＤＥ_０〜ＮＯＤＥ_ｎに遷移するかを各ノードＮＯＤＥ_０〜ＮＯＤＥ_ｎ間を接続するアークＡＲＣ_０〜ＡＲＣ_ｎに対してそれぞれ設定された遷移確率Ｐ_０〜Ｐ_ｎに基づいて確率的に決定する確率オートマトンと呼ばれるアルゴリズムを用いる。
【００６１】
より具体的には、メモリ１０Ａには、各行動モデルごとに、その行動モデルの各ノードＮＯＤＥ_０〜ＮＯＤＥ_ｎごとの図６に示すような状態遷移表３５が格納されている。
【００６２】
ここで状態遷移表３５では、そのノードＮＯＤＥ_０〜ＮＯＤＥ_ｎにおいて遷移条件とする入力イベント（状態認識部３０の認識結果）が「入力イベント」の行に優先順に列記され、その条件についてのさらなる条件が「データ名」及び「データ範囲」の行における対応する列に記述されている。
【００６３】
従って図６の状態遷移表３５で定義されたノードＮＯＤＥ_１００では、「ボールを検出した（ＢＡＬＬ）」という認識結果が与えられた場合に、当該認識結果と共に与えられるそのボールの「大きさ（ＳＩＺＥ）」が「０から１００の範囲（０，１０００）」であることや、「障害物を検出（ＯＢＳＴＡＣＬＥ）」という認識結果が与えられた場合に、当該認識結果と共に与えられるその障害物までの「距離（ＤＩＳＴＡＮＣＥ）」が「０から１０００の範囲（０，１０００）」であることが自己又は他のノードＮＯＤＥ_０〜ＮＯＤＥ_ｎに遷移するための条件となる。
【００６４】
またこのノードＮＯＤＥ_１００では、認識結果の入力がない場合においても、行動決定部３２が周期的に参照する感情・本能モデル部３１の各情動及び各欲求のパラメータ値のうち「喜び（ＪＯＹ）」、「驚き（ＳＵＰＲＩＳＥ）」又は「悲しみ（ＳＵＤＮＥＳＳ）」のいずれかの情動のパラメータ値が「５０から１００の範囲（５０，１００）」であるときには自己又は他のノードＮＯＤＥ_０〜ＮＯＤＥ_ｎに遷移することができる。
【００６５】
さらに状態遷移表３５では、「他のノードへの遷移確率」の欄における「遷移先ノード」の列にそのノードＮＯＤＥ_０〜ＮＯＤＥ_ｎから遷移できるいくつかのノードＮＯＤＥ_０〜ＮＯＤＥ_ｎが列記されると共に、「入力イベント名」、「データ値」及び「データの範囲」の各行に記述された全ての条件が揃った場合におけるそのノードＮＯＤＥ_０〜ＮＯＤＥ_ｎへの遷移確率が「他のノードへの遷移確率」の欄におけるそのノードＮＯＤＥ_０〜ＮＯＤＥ_ｎの行に記述され、このとき出力される行動及び動作が「出力行動」の行に記述される。なお「他のノードへの遷移確率」の欄における各行の遷移確率の和は１００ 〔％〕となっている。
【００６６】
従ってこの例のノードＮＯＤＥ_１００では、例えば「ボールを検出（ＢＡＬＬ）」し、そのボールの「大きさ（ＳＩＺＥ）」が「０から１０００の範囲（０，１０００）」であるという認識結果が与えられた場合には、「３０〔％〕」の確率で「ノードＮＯＤＥ_１２０（ｎｏｄｅ１２０）」に遷移でき、そのとき「ＡＣＴＩＯＮ １」の行動又は動作が出力されることとなる。
【００６７】
そして各行動モデルは、それぞれこのような状態遷移表３５として記述されたノードＮＯＤＥ_０〜ＮＯＤＥ_ｎがいくつも繋がるようにして形成されている。
【００６８】
かくして行動決定部３２は、状態認識部３０から状態認識情報Ｓ１０が与えられたときや、最後に行動を発現してから一定時間が経過したときなどに、メモリ１０Ａに格納されている対応する行動モデルのうちの対応するノードＮＯＤＥ_０〜ＮＯＤＥ_ｎの状態遷移表３５を利用して次の行動や動作（「出力行動」の行に記述された行動又は動作）を確率的に決定し、決定結果を行動決定情報Ｓ１１として感情・本能モデル部３２、行動生成部３３及び成長制御部３４に出力する。
【００６９】
（２−４）行動生成部３３の構成
行動生成部３３は、各「成長段階」にそれぞれ対応させて、その「成長段階」に応じた「成長レベル」の「歩行状態」、「モーション」及び「サウンド」等の行動や動作をペットロボット１に発現させるための各種制御パラメータをメモリ１０Ａ内に有している。
【００７０】
そして行動生成部３３は、行動決定部３２から行動決定情報Ｓ１１が与えられると、メモリ１０Ａに格納されているこれら各種制御パラメータのうち、後述のように予め成長制御部３４により指定された「成長段階」の制御パラメータに基づいて必要なアクチュエータ２１_１〜２２_ｎを所定状態に駆動させたり、音声信号Ｓ３に基づく音声をスピーカ２０から出力させたり、外見上の「目」としてのＬＥＤ１９を所定パターンで点滅駆動する。
【００７１】
このようにして行動生成部３３は、行動決定部３２により決定された行動を、そのときの「成長段階」に応じたレベルでペットロボット１に発現させ得るようになされている。
【００７２】
（２−５）成長制御部３４の構成
成長制御部３４は、図７（Ａ）に示すように、状態認識部３０から与えられる状態認識情報Ｓ１０に基づく各種状態のうち、「成長段階」を上げる際の参考要素とすべき上述の成長要素のリスト（以下、これを第１の成長要素リストと呼ぶ）４０Ａと、これら成長要素の累積度数をそれぞれ計数するための図７（Ｂ）のようなカウンタテーブル（以下、これを第１の成長要素カウンタテーブルと呼ぶ）４１Ａとをメモリ１０Ａ内に有している。
【００７３】
そして成長制御部３４は、状態認識部３０から状態認識情報Ｓ１０が与えられると、当該状態認識情報Ｓ１０に基づき得られる状態が成長要素か否かを第１の成長要素リスト４０Ａに基づいて判断し、当該状態が成長要素である場合には第１の成長要素カウンタテーブル４１Ａ内の対応するカウント値（経験値）を１つ増加させる。
【００７４】
また成長制御部３４は、図８（Ａ）に示すように、上述のように行動決定部３２から与えられる行動決定情報Ｓ１１に基づき得られる行動のうち、「成長段階」を上げる際の参考要素とすべき上述の成長要素のリスト（以下、これを第２の成長要素リストと呼ぶ）４０Ｂと、これら成長要素の累積度数をそれぞれ計数するための図８（Ｂ）のようなカウンタテーブル（以下、これを第２の成長要素カウンタテーブルと呼ぶ）４１Ｂとをメモリ１０Ａ内に有している。
【００７５】
そして成長制御部３４は、行動決定部３２から行動決定情報Ｓ１１が与えられると、当該行動決定情報Ｓ１１に基づき得られる行動が成長要素か否かを第２の成長要素リスト４０Ｂに基づいて判断し、当該行動が成長要素である場合には第２の成長要素カウンタテーブル４１Ｂ内の対応するカウント値（経験値）を１つ増加させる。
【００７６】
さらに成長制御部３４は、上述のように第１又は第２の成長要素カウンタテーブル４１Ａ、４１Ｂ内のカウント値を増加させたときには、これら第１及び第２の成長要素カウンタテーブル４１Ａ、４１Ｂとは別に用意した「成長段階」を上げるか否かを判定するためのカウンタ（以下、これを総合経験値カウンタと呼ぶ）のカウント値を１増加させ、この後当該総合経験値カウンタのカウント値が現在の「成長段階」の終了条件として予め設定されたカウント値に達したか否かを判断する。
【００７７】
そして成長制御部３４は、総合経験値カウンタのカウント値が現在の「成長段階」の終了条件として予め設定されたカウント値に達した場合には、ペットロボット１の「性格」を次の「成長段階」におけるどの「性格」とするかを第１及び第２の成長要素カウンタテーブル４１Ａ、４１Ｂ内の各カウント値に基づいて決定し、決定結果を性格指定情報Ｓ２０として行動決定部３２及び行動生成部３３に出力する。なお成長制御部３４は、初期時には「幼年期」の「性格」（Ｂａｂｙ １）を選択すべき旨の性格指定情報Ｓ２０を行動決定部３２及び行動生成部３３に与える。
【００７８】
この結果行動決定部３２は、かかる性格指定情報Ｓ２０に基づいて、メモリ１０Ａに格納されている複数の行動モデルの中から指定された「性格」と対応付けられた行動モデルを選択し、これ以降はこの行動モデルを用いて上述のように現在のペットロボット１の行動に対する次の行動を決定するようになる。
【００７９】
また行動生成部３３は、かかる性格指定情報Ｓ２０に基づいて、指定された「成長段階」と対応付けられた「歩行状態」、「モーション」及び「サウンド」等についての制御パラメータを選択し、これ以降はこの制御パラメータを用いて各アクチュエータ２１_１〜２１_ｎ等を駆動する。
【００８０】
このようにして成長制御部３４は、ペットロボット１の「成長」を制御し得るようになされている。
【００８１】
（３）行動決定部３２の詳細構成
次にこのペットロボット１の行動決定部３２の詳細構成について説明する。
【００８２】
このペットロボット１では、当該ペットロボット１の行動をその内容に応じて図９のように「システム系」、「転倒復帰系」、「反応系」、「障害物回避系」、「感情がやや高いとき系」、「基本行動系」などの幾つかの系統（以下、これらを主分類系統と呼ぶ）５０〜５５、……に分け、これら各主分類系統５０〜５５、……ごとに、それぞれペットロボット１の「成長段階」や「性格」に応じた行動を発現し得るように１又は複数の行動モデルを用意している。
【００８３】
またこのペットロボット１では、自律モード時における当該ペットロボット１の状態を、図１０のようにしばらくの間ユーザからの入力がないときなどの状態である「退屈状態」と、ユーザやボールと遊んでいるときなどの状態である「わんぱく状態」、ボール等を探すなどして歩き回っている「お散歩状態」と、退屈そうにごろごろと寝ころがっているときなどの状態である「ぐうたら状態」と、座った姿勢でユーザと遊んでいる「おはなし／おしえて状態」と、寝ているときなどの状態である「おやすみ状態」の６つの基本状態に分け、これら６つの基本状態のうちの行動が発現されない「おやすみ状態」を除く残りの５つの基本状態にそれぞれ対応させて、「基本行動系」の主分類系統５５を「退屈系」、「追跡系」、「歩き回り系」、「狸寝入り系」及び「インタラクション系」の５つの系統（以下、この系統を副分類系統と呼ぶ）５５Ａ〜５５Ｅに分け、これら各副分類系統５５Ａ〜５５Ｅごとにそれぞれペットロボット１の「性格」に応じた行動を発現し得るように複数の行動モデルを用意している。
【００８４】
そして行動決定部３２は、例えば成長制御部３４から上述の性格指定情報Ｓ２０が与えられた場合には、２以上の行動モデルが用意されている「基本行動系」以外の各主分類系統５０〜５４について、性格指定情報Ｓ２０により指定された「性格」の行動を発現し得るように、予めその「性格」に対応付けられている行動モデルを１つ選択する。
【００８５】
例えばこの図９の例では、指定された「性格」が幼児期の「おっとりとした性格」（Ｃｈｉｌｄ １）（図３）であったことから、「転倒復帰系」については転倒したときに復帰できずに呼ぶ又は失神するという行動を発現し易い「復帰できずに呼ぶ／失神」という行動モデルが選択され、「障害物回避系」については障害物に対して近寄らない行動を発現し易い「チキン（臆病者）」という行動モデルが選択されていることを示している。なお「システム系」、「反応系」及び「感情がやや高いとき系」については、もともと行動モデルが１つしか用意されていないため、それぞれ「Ｐｌａｎｅ（平常）」という行動モデルが選択される。
【００８６】
またこれと共に行動決定部３２は、「基本行動系」の主分類系統５５に関しても、当該「基本行動系」内の各副分類系統５５Ａ〜５５Ｅについて、それぞれ性格指定情報Ｓ２０により指定された「性格」の行動を発現し得るように、予めその「性格」に対応付けられている行動モデルを１つずつ選択する。
【００８７】
例えば図９に示す例では、「退屈系」については「きょろ系」、「追跡系」については「ユーザ系」、「歩き回り系」については「ユーザ探索場所移動系」、「狸寝入り系」については「すやすや系」、「インタラクション系」については「すなお系」の行動モデルがそれぞれ選択されたことを示している。
【００８８】
そして行動決定部３２は、この後、状態認識部３０から状態認識情報Ｓ１０が与えられたときや、最後に行動を発現してから一定時間が経過したときなどに、「基本行動系」を除く主分類系統５０〜５４のうちの対応する全ての主分類系統５０〜５４の行動モデルをそれぞれ用いて次の行動をそれぞれ決定する。
【００８９】
なおここでいう『対応する全ての主分類系統５０〜５４』とは、状態認識部３０からの認識結果等に対して予め対応付けられている主分類系統５０〜５４のことを意味し、例えば状態認識部３０からの「内部温度が所定温度以上となった」や「バッテリ残量が少なくなった」という認識結果に対しては「システム系」のみが対応付けられ、「転倒した」という認識結果に対しては「転倒復帰系」のみが対応付けられている。また最後に行動を発現してから一定時間が経過した場合に対しては、「感情がやや高いとき系」及び「基本行動系」が対応付けられている。従ってこの場合には、「感情がやや高いとき系」及び「基本行動系」のそれぞれの行動モデルを用いて続く行動がそれぞれ決定される。
【００９０】
またこれと共に行動決定部３２は、「基本行動系」については、状態認識部３０からの認識結果等に対して予め対応付けられている１つの副分類系統５５Ａ〜５５Ｅの行動モデルを用いて続く行動を決定する。例えば状態認識部３０からの「ボールを検出した」との認識結果には、「追跡系」が対応付けられており、従ってこの場合には「追跡系」の行動モデルを用いて続く行動が決定される。
【００９１】
そして行動決定部３２は、上述のようにして対応する全ての主分類系統５０〜５４の各行動モデルや、対応する副分類系統５５Ａ〜５５Ｅ内の１つの行動モデルを用いて次の行動を決定すると、この後これら決定した行動の中から予め定められた優先順位の高い主分類系統５０〜５５によって選択された行動を１つ選択し、当該選択した行動を上述のように行動決定情報Ｓ１１として行動生成部３２及び成長制御部３４に通知する。因にこの実施の形態においては、図９において上側に表記された主分類系統５０〜５５ほど優先順位が高く設定されている。
【００９２】
このようにしてこの行動決定部３２は、そのときのペットロボット１の「性格」に応じた行動モデルを利用して続く行動を決定し得るようになされている。
【００９３】
なおこの実施の形態においては、同じ「成長段階」の同じ「性格」のペットロボット１でも、それまでの外部からの入力履歴や自己の行動履歴に応じて、出力行動の難易度や煩雑さに変化を生じさせ得るようになされている。
【００９４】
すなわちこのペットロボット１の場合、行動決定部３２は、図１１に示すように、各種芸や、大きな音に対する反応（ＬＯＵＤ）、動きの難易度（ＭＯＶＥ）、ボールに対する反応（ＢＡＬＬ）及び好きな色に対する反応（ＦＡＶ）などの幾つかの行動や動作について、同じ行動や動作でも難易度や煩雑さ等が異なる幾つかのパターンを有している。
【００９５】
そして成長制御部３４は、上述のようにペットロボット１の「成長段階」を上げる際に第１及び第２の成長要素カウンタテーブル４１Ａ、４１Ｂ（図７（Ｂ）、図８（Ｂ））内の各カウント値に基づいて、かかる行動や動作についてそれぞれどのパターンの行動や動作を使用すべきかを行動決定部３２に通知（以下、これを行動難易度指定通知と呼ぶ）する。例えばこの図１１は「性格」が幼児期の「おっとりとした性格」（Ｃｈｉｌｄ １）（図３）である場合の一例を示すのもであり、この例では出力される各種芸については「稚拙」、大きな音に対する反応については「失神」、ボールに対する接触の仕方については「稚拙」、お手の仕方については「稚拙」が指定されたことを表す。
【００９６】
また行動決定部３２は、かかる行動難易度指定通知に基づいて、指定された行動や動作のパターンを選択し、かかる行動や動作についてそのパターンが発現されるように、図９に示す「基本行動系」に属する全ての行動モデル（「きょろ系」、「のんびり系」、……、「すやすや系」、「下品系」、……、「いたずら系」、「すなお系」、「独立系」、……、「ＦＡＶ定住場所微少系」、「ボール探索場所移動系」、「ユーザ探索場所移動系」、……、「アイテム系」、「ユーザ系」、……）における対応する状態遷移表３５（図６）内の対応する出力行動の欄を対応する出力行動に変更する。
【００９７】
この結果「基本行動系」に属する行動モデルを用いて次の行動としてかかる行動や動作が選択されたときに、成長制御部３４により指定されたパターンの行動や動作が発現されることとなる。
【００９８】
このようにしてこのペットロボット１では、同じ「成長段階」の同じ「性格」のものであっても、各行動や動作ごとにそれぞれ異なるパターン（難易度・煩雑さ）の行動や動作を発現し得るようにすることができ、これによりペットロボット１に多様な個性をもたせ得るようになされている。
【００９９】
（４）本実施の形態の動作及び効果
以上の構成において、このペットロボット１では、自律モード時における当該ペットロボット１の状態を複数の基本状態に分けたときの各基本状態ごとの行動モデル（図９の「基本状態系」の「退屈系」、「追跡系」、「歩き回り系」、「狸寝入り系」又は「インタラクション系」に属する各行動モデル）を用意し、これら行動モデルのうち、状態認識部３０の認識結果等に応じた行動モデルを用いて行動決定部３２において行動を決定し、当該決定した行動を発現するようにアクチュエータ２１_１〜２１_ｎ等を制御する。
【０１００】
従ってこのペットロボット１では、それまで１つであった「基本状態系」の行動モデルの規模を縮小させてその取扱いを容易化させることができ、また基本状態の分け方によっては行動モデルの数を増加させることができるため、ペットロボット１の行動様式を多様化させることができる。
【０１０１】
以上の構成によれば、自律モード時における当該ペットロボット１の状態を複数の基本状態に分けたときの各基本状態ごとの行動モデル（図９の「基本状態系」の「退屈系」、「追跡系」、「歩き回り系」、「狸寝入り系」又は「インタラクション系」に属する各行動モデル）を用意し、これら行動モデルのうち、状態認識部３０の認識結果等に応じた行動モデルを用いて行動決定部３２において行動を決定し、当該決定した行動を発現するようにアクチュエータ２１_１〜２１_ｎ等を制御するようにしたことにより、行動モデルの規模を縮小させてその取扱いを容易化させることができると共に、基本状態の分け方によっては行動モデルの数を増加させることができることからペットロボット１の行動様式を多様化させることができ、かくして行動様式の変更及び多様化を容易化し得るペットロボットを実現できる。
【０１０２】
（５）他の実施の形態
なお上述の実施の形態においては、本発明を図１のように構成された４足歩行型のペットロボットに適用するようにした場合について述べたが、本発明はこれに限らず、要は、外部センサ及び又は内部センサの出力に基づいて外部及び又は内部の状態を認識し、認識結果に基づいて行動する自律型のロボット装置であるのならば、この他種々の形態のロボット装置に広く適用することができる。
【０１０３】
また上述の実施の形態においては、自律モード時におけるペットロボット１のの基本状態を「退屈状態」、「わんぱく状態」、「お散歩状態」、「ぐうたら状態」、「おはなし／おしえて状態」及び「おやすみ状態」の６つの状態に分けるようにした場合について述べたが、本発明はこれに限らず、この他種々の状態に分けるようにしても欲、また６以外の状態に分けるようにしても良い。
【０１０４】
さらに上述の実施の形態においては、各種行動モデルを記憶する記憶手段としてメモリ１０Ａを適用するようにした場合について述べたが、本発明はこれに限らず、これ以外の記憶手段を適用するようにしても良い。
【０１０５】
さらに上述の実施の形態においては、メモリ１０Ａに格納された各種行動モデルのうち、外部及び又は内部の状況に応じた行動モデルを用いて行動を決定する行動決定手段としての行動決定部３２と、行動決定部３２により決定された行動を発現するように所定の制御処理を実行する制御手段としての行動生成部３３と、外部からの入力履歴及び又は自己の行動履歴に応じて、上記記憶手段に記憶された各上記行動モデルを変更する変更手段としての成長制御部３４とを同じコントローラ１０により構成するようにした場合について述べたが、本発明はこれに限らず、これらを別体として構成するようにしても良い。
【０１０６】
さらに上述の実施の形態においては、ペットロボット１の「成長段階」を上げる際に図９に示す「基本行動系」に属する各行動モデルの内容を変更するようにした場合について述べたが、本発明はこれに限らず、そのタイミングとしては「成長段階」を上げる以外のタイミングであっても良い。この場合においては、例えば外部からの入力履歴及び又は自己の行動履歴に基づく学習によりそのタイミングを設定するようにすれば良い。
【０１０７】
【発明の効果】
上述のように本発明によれば、外部センサ及び内部センサのうち少なくともいずれか一方の出力に基づいて外部及び内部のうち少なくともいずれか一方の状況を認識し、外部及び内部のうち少なくともいずれか一方の状況のうち、予め定められた成長に関する複数の成長要素の累積度を当該成長要素ごとにカウントするカウンタと、成長要素の累積度の合計に応じ、次の成長段階へ移行すると共に、行動パターンの系列である基本行動系が定められた複数の性格から成長要素の累積度に応じた性格を選択し、成長要素の累積度に応じて基本行動系に属する行動パターンに対する難易度を設定する成長制御部と、外部及び内部のうち少なくともいずれか一方の状況に対し、選択された性格における基本行動系から、設定された難易度でなる行動パターンを選択する行動決定部と、行動決定部によって選択された行動パターンに従って行動するよう、駆動部を制御する駆動制御部とを設けるようにしたことにより、行動パターンのバリエーションを増加させることができ、かくして行動様式の変更及び多様化を容易かし得るロボット装置を実現できる。
【０１０８】
また本発明によれば、外部センサ及び内部センサのうち少なくともいずれか一方の出力に基づいて外部及び内部のうち少なくともいずれか一方の状況を認識し、外部及び内部のうち少なくともいずれか一方の状況のうち、予め定められた成長に関する複数の成長要素の累積度を当該成長要素ごとにカウントするカウンタステップと、成長要素の累積度の合計に応じ、次の成長段階へ移行すると共に、行動パターンの系列である基本行動系が定められた複数の性格から成長要素の累積度に応じた性格を選択し、成長要素の累積度に応じて基本行動系に属する行動パターンに対する難易度を設定する成長制御ステップと、外部及び内部のうち少なくともいずれか一方の状況に対し、選択された性格における基本行動系から、設定された難易度でなる行動パターンを選択する行動決定ステップと、行動決定部によって選択された行動パターンに従って行動するよう、駆動部を制御する駆動制御ステップとを設けるようにしたことにより、行動パターンのバリエーションを増加させることができ、かくして行動様式の変更及び多様化を容易かし得るロボット装置を実現できる。
【図面の簡単な説明】
【図１】本実施の形態によるペットロボットの外観構成を示す斜視図である。
【図２】本実施の形態によるペットロボットの内部構成を示すブロック図である。
【図３】ペットロボットの「成長」の説明に供する概念図である。
【図４】コントローラの処理の説明に供するブロック図である。
【図５】確率オートマトンを示す概念図である。
【図６】状態遷移表を示す概念図である。
【図７】第１の成長要素リスト及び第１の成長要素カウンタテーブルの説明に供する概念図である。
【図８】第２の成長要素リスト及び第２の成長要素カウンタテーブルの説明に供する概念図である。
【図９】行動決定部の説明に供する概念図である。
【図１０】自律モード時におけるペットロボットの各種基本状態の説明に供する略線図である。
【図１１】自律モード時におけるペットロボットの各種基本状態の説明に供する略線図である。
【符号の説明】
１……ペットロボット、１０……コントローラ、１０Ａ……メモリ、３０……状態認識部、３１……感情・本能モデル部、３２……行動決定部、３３……行動生成部、３４……成長制御部、５０〜５５……主分類系統、５５Ａ〜５５Ｅ……副分類系統、Ｓ１０……状態認識情報、Ｓ１１……行動決定情報、Ｓ２０……性格指定情報。[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a robot apparatus and a control method thereof, and is suitably applied to, for example, a pet robot.
[0002]
[Prior art]
In recent years, a pet robot designed to behave autonomously according to an instruction from a user and the surrounding environment has been developed by the present applicant. Such a pet robot has an appearance resembling a dog or a cat bred in a general home, and is based on the output of a CCD (Charge Coupled Device) camera, a microphone, a touch sensor, and the like disposed in each predetermined part. It recognizes the internal situation, the command from the user, the presence or absence of an action, and the like, and acts based on the recognition result.
[0003]
[Problems to be solved by the invention]
By the way, in such a conventional pet robot, it is constructed so as to determine an action to be expressed by using a single model that defines a behavior mode called a behavior model. There was a problem that was difficult to diversify.
[0004]
The present invention has been made in view of the above points, and an object of the present invention is to propose a robot apparatus and a control method thereof that can facilitate the change and diversification of behavior patterns.
[0005]
[Means for Solving the Problems]
In order to solve this problem, in the present invention, At least one of external sensor and internal sensor Based on the output of At least one of the external and internal situations Recognize In addition, according to a counter that counts the cumulative degree of a plurality of growth factors related to a predetermined growth in each of the growth factors in at least one of the external and internal situations, While moving to the next growth stage, select a personality according to the cumulative degree of growth elements from a plurality of personalities with basic behavioral systems that are a series of behavior patterns, and select basic behavioral systems according to the cumulative degree of growth elements Select the behavior pattern with the set difficulty level from the basic behavior system with the selected personality for the growth control unit that sets the difficulty level for the behavior pattern belonging to and at least one of the external and internal situations A behavior control unit that controls the drive unit to act according to the behavior pattern selected by the behavior determination unit; It was made to provide.
[0006]
As a result, this robotic device ,line Movement pattern The variation of can be increased.
[0007]
In the present invention, At least one of external sensor and internal sensor Based on the output of At least one of the external and internal situations Recognize In addition, a counter step for counting the cumulative degree of a plurality of growth factors relating to a predetermined growth in at least one of the external and internal situations for each growth factor, and depending on the total cumulative factor of the growth factors In addition to moving to the next growth stage, select a personality according to the cumulative degree of growth factors from a plurality of personalities with a basic behavior system that is a series of behavior patterns, and select basic behavior according to the cumulative degree of growth factors A growth control step for setting a difficulty level for an action pattern belonging to the system, and an action pattern having a set difficulty level from the basic action system for the selected character for at least one of the external and internal situations. An action determining step to select, and a drive control step to control the drive unit to act according to the action pattern selected by the action determining unit; It was made to provide.
[0008]
As a result, this robotic device ,line Movement pattern The variation of can be increased.
[0009]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, an embodiment of the present invention will be described in detail with reference to the drawings.
[0010]
(1) Configuration of pet robot 1 according to the present embodiment
In FIG. 1, 1 indicates a pet robot according to the present embodiment as a whole, and leg units 3 </ b> A to 3 </ b> D are connected to the front, rear, left, and right of the body unit 2, respectively, and the front and rear ends of the body unit 2. The head unit 4 and the tail unit 5 are connected to each other.
[0011]
In this case, as shown in FIG. 2, the body unit 2 includes a controller 10 that controls the behavior of the entire pet robot 1, a battery 11 as a power source of the pet robot 1, a battery sensor 12, and a temperature sensor 13. The internal sensor unit 14 and the like made up of and the like are housed.
[0012]
The head unit 4 includes an external sensor unit 18 including a CCD (Charge Coupled Device) camera 15 that functions as a substantial “eye” of the pet robot 1, a microphone 16 that functions as an “ear”, a touch sensor 17, and the like. In addition, an LED (Light Emitting Diode) 19 that functions as an “eye” in appearance, a speaker 20 that functions as a “mouth”, and the like are disposed at predetermined positions.
[0013]
Further, the joint portions of the leg units 3A to 3D, the connecting portions of the leg units 3A to 3D and the torso unit 2, the connecting portions of the head unit 4 and the torso unit 2, and the tail in the tail unit 5 For the root portion of 5A (FIG. 1), etc., there are actuators 21 for the required number of degrees of freedom. ₁ ~ 21 _n Is arranged.
[0014]
Then, the CCD camera 15 of the external sensor unit 18 captures the surrounding situation and sends the obtained image signal S1A to the controller 10. Further, the microphone 16 collects a command sound such as “walk”, “turn down”, or “follow the ball” given as a musical scale via a sound commander (not shown) from the user, and the obtained audio signal S1B is sent to the controller 10. Send it out.
[0015]
Further, as apparent from FIG. 1, the touch sensor 17 is provided on the upper part of the head unit 4, and detects a pressure received by a physical action such as “blowing” or “striking” from the user. Is sent to the controller 10 as a pressure detection signal S1C.
[0016]
On the other hand, the battery sensor 12 of the internal sensor unit 14 detects the remaining amount of the battery 11, sends the detection result to the controller 10 as a battery remaining amount detection signal S2A, and the temperature sensor 13 determines the internal temperature of the pet robot 1. The detected result is sent to the controller 10 as a temperature detection signal S2B.
[0017]
The controller 10 includes an image signal 1A, an audio signal S1B, and a pressure detection signal S1C (hereinafter collectively referred to as an external sensor signal S1) supplied from the CCD camera 15, the microphone 16, and the touch sensor 17, and a battery sensor 12. And the situation around and inside the pet robot 1 based on the battery remaining amount detection signal S2A and the temperature detection signal S2B (hereinafter collectively referred to as the internal sensor signal S2) given from the temperature sensor 13, respectively, It judges the command from the user, the presence / absence of the action from the user, and the like.
[0018]
Then, the controller 10 determines the action to be continued based on the determination result and the control program stored in the memory 10A in advance, and the necessary actuator 21 based on the determination result. ₁ ~ 21 _n , The head unit 4 is swung up and down, left and right, the tail 5A of the tail unit 5 can be moved, and the leg units 3A to 3D are driven to walk.
[0019]
At this time, the controller 10 outputs a sound based on the sound signal S3 to the outside by giving a predetermined sound signal S3 to the speaker 20 as necessary, or causes the LED 19 as an “eye” to appear as an LED drive signal. This is blinked by outputting S4.
[0020]
In this way, the pet robot 1 can act autonomously based on the surrounding or internal situation, the command from the user, the presence or absence of an action, and the like.
[0021]
In addition to such a configuration, in the case of this pet robot 1, a real animal will “grow” according to the history of operation inputs such as user's actions and commands using a sound commander, and its own behavior and movement history. It is designed to change behavior and movement as if it were.
[0022]
That is, as shown in FIG. 3, the pet robot 1 is provided with four “growth stages” of “childhood”, “boyhood”, “adolescence”, and “adulthood” as growth processes. The memory 10A of the controller 10 stores in advance control parameters that are the basis of actions and operations such as “walking state”, “motion” and “sound” for each “growth stage”. ing.
[0023]
Then, the controller 10 simply “walks” and “walks” for “motion” so that the “walking state” becomes “walking tow”, for example, by reducing the stride in the “walking state” in accordance with the control parameter of “childhood” at the initial stage. Each actuator 21 has a “simple” motion such as “standing” and “sleeping”, and “sound” is made “small and short” by reducing the amplification factor of the audio signal S6. ₁ ~ 21 _n And control the audio output.
[0024]
At this time, the controller 10 also performs reinforcement learning including command input using the sound commander, sensor input via the touch sensor 17 corresponding to “stroking” and “tapping”, and the number of successful actions and actions determined. And a plurality of sensors that are involved in predetermined “growth” such as sensor input via the touch sensor 17 that does not correspond to “stroking” and “tapping”, and predetermined actions and actions such as “playing with a ball”. The occurrence of elements (hereinafter referred to as growth elements) is constantly monitored and counted.
[0025]
Based on the cumulative frequencies of these growth factors, the controller 10 uses when the total value of the cumulative frequencies of each growth factor (hereinafter referred to as the total experience value of the growth factors) exceeds a preset threshold value. The control parameter is changed to a control parameter of “boyhood” having a higher growth level (level of difficulty or complexity of behavior and action) than “childhood”.
[0026]
Then, the controller 10 thereafter controls each actuator 21 for the “walking state” in accordance with the control parameter of the “boyhood”, for example. ₁ ~ 21 _n Increase the number of movements to increase the number of movements, such as increasing the number of movements, and increase the number of movements. For each of the actuators 21, the amplification factor of the audio signal S 6 is increased to make a “slightly longer and louder” call. ₁ ~ 21 _n And the sound output from the speaker 20 is controlled.
[0027]
Further, the controller 10 thereafter increases the control parameter every time the total experience value of the growth factor exceeds each threshold set in advance corresponding to “adolescence” and “adulthood”. The behavior and operation model of “adolescence” or “adult” at a high level is sequentially changed, and each actuator 21 is changed according to the control parameter. ₁ ~ 21 _n Each actuator 21 for gradually increasing the rotation speed, the length of the audio signal S3 applied to the speaker 20 and the amplification factor, or performing one operation. ₁ ~ 21 _n Change the amount of rotation.
[0028]
As a result, the pet robot 1 increases in the “growth stage” (ie, changes from “childhood” to “boyhood”, “boyhood” to “adolescence”, “adolescence” to “adulthood”), "Walking state" gradually changes from "Toddler to walk" to "Solid walking", "Motion" from "Simple" to "Advanced / Complex", and "Sound" from "Small and short" to "Long and large" Change.
[0029]
In this way, the pet robot 1 has four stages of “childhood”, “boyhood”, “adolescence”, and “adulthood” according to the input from the outside and the history of self-action and movement. It is designed to “grow”.
[0030]
In the case of this embodiment, as is apparent from FIG. 3, a plurality of “characters” are prepared for each “growth stage” of “boyhood”, “adolescence”, and “adulthood”.
[0031]
In practice, for example, “boyhood” “personality” includes “rough personality” (Child 1), and “moist personality” (Child 2) that is smoother and slower than this. is there.
[0032]
Also, as the “personality” of “Youth”, “Iraira character” (Young 1), which is slower and smoother than the “rough personality” of “Youth”, the slower and smoother “ There are “ordinary personality” (Young 2) and “soft character” (Young 3) that is slower than this and has a small amount of action.
[0033]
Furthermore, as the “personality” of “adulthood”, the “aggressive personality” (which is more complicated and faster than the “irritated personality” of “adolescence” and difficult to move in response to commands from the user ( “Adult 1)” and “Slightly rough personality” (Adult 2) that makes the movement smoother and slower than this, and can easily move according to the command from the user. “Slightly gentle personality” (Adult 3) that always moves according to the command from the user, and this makes the movement even slower, the amount of action is small, and the movement according to the command from the user There is always a “gentle character” (Adult 4).
[0034]
Then, when increasing the “growth stage”, the controller 10 selects one “character” in the next “growth stage” based on the cumulative frequency of each growth factor.
[0035]
In this case, after “boyhood”, when moving to the next “growth stage”, the “character” of the next “growth stage” that can be transitioned from the “character” of the current “growth stage” has been determined. You can only transition between “characters” connected by arrows. Therefore, for example, when “rough personality” (Child 1) is selected in “boyhood”, it is not possible to transition to “moist personality” (Young 3) in “adolescence”.
[0036]
In this way, in this pet robot, according to the action from the user, the command input history, and the self action history, as if a real animal forms a personality according to the way of breeding the owner, etc. The “personality” is changed along with the “growth”. A method for causing the pet robot 1 to perform an action or an action according to the selected “personality” will be described later.
[0037]
(2) Processing of controller 10
Next, specific processing of the controller 10 in the pet robot 1 will be described.
[0038]
The controller 10 executes various processes according to the control program stored in the memory 10A (FIG. 2). Here, when the processing contents of the controller 10 are functionally classified, as shown in FIG. 4, the external and internal states are based on the outputs of the external sensor unit 18 (FIG. 2) and the internal sensor unit 14 (FIG. 2). The state recognition unit 30 for recognizing the emotion, the emotion / instinct model unit 31 for determining the emotion and instinct state of the pet robot 1 based on the recognition result of the state recognition unit 30, the recognition result of the state recognition unit 30, and the pet robot A behavior determination unit 32 that determines the subsequent behavior based on the state of one emotion and instinct, a behavior generation unit 33 that causes the pet robot 1 to express a behavior according to the determination result of the behavior determination unit 32, and “ It can be divided into a growth control unit 34 for controlling “growth”.
[0039]
Hereinafter, the state recognition unit 30, the emotion / instinct model unit 31, the behavior determination unit 32, the behavior generation unit 33, and the growth control unit 34 will be described in detail.
[0040]
(2-1) Configuration of state recognition unit 30
The state recognition unit 30 recognizes a specific state based on the external sensor signal S1 provided from the external sensor unit 18 and the internal sensor signal S2 provided from the internal sensor unit 14, and uses the recognition result as the state recognition information S10 as an emotion. Notify the instinct model unit 31 and the action determination unit 32.
[0041]
In practice, the state recognizing unit 30 constantly monitors the image signal S1A given from the CCD camera 15 (FIG. 2) of the external sensor unit 18, and the image based on the image signal S1A includes, for example, “red round object” and “vertical”. When the “smooth plane” is detected, it recognizes that “the ball is present” and “there is a wall”, and notifies the emotion / instinct model unit 31 and the action determination unit 32 of the recognition result.
[0042]
In addition, the state recognition unit 30 constantly monitors the audio signal S1B given from the microphone 16 (FIG. 2), and receives command sounds such as “walk”, “down” and “follow the ball” based on the audio signal S1B. When it is recognized, the recognition result is notified to the emotion / instinct model unit 31 and the action determination unit 32.
[0043]
Further, the state recognition unit 30 constantly monitors the pressure detection signal S1C given from the touch sensor 17 (FIG. 2), and based on the pressure detection signal S1C, the pressure for a short time (for example, less than 2 seconds) is greater than a predetermined threshold value. It is recognized as “struck (struck)” when it is detected, and “struck (admired)” when a wide range of pressures is detected for a long time (eg, 2 seconds or more) below a predetermined threshold. Recognize and notify the emotion / instinct model unit 31 and the action determination unit 32 of the recognition result.
[0044]
Further, the state recognizing unit 30 constantly monitors the temperature detection signal S2B given from the temperature sensor 13 (FIG. 2) of the internal sensor unit 14, and when detecting a temperature higher than a predetermined value based on the temperature detection signal S2B, the “internal temperature” And the recognition result is notified to the emotion / instinct model unit 31 and the action determination unit 32.
[0045]
(2-2) Configuration of Emotion / Instinct Model Unit 31
The emotion / instinct model unit 31 is a parameter that represents the strength of emotion for each of the six emotions of “joy”, “sadness”, “surprise”, “fear”, “disgust”, and “anger”. Holding. Then, the emotion / instinct model unit 31 sets the parameter value of each emotion as a specific recognition result such as “struck” or “boiled” given as state recognition information S10 from the state recognition unit 30, respectively. As described above, the action determination information S11 representing the determined output action given from the action determination unit 32 and the elapsed time are sequentially updated.
[0046]
Specifically, the emotion / instinct model unit 31 determines whether the recognition result obtained based on the state recognition information S10 and the output action based on the action determination information S11 act on the emotion (set in advance), Recognize the amount of change in the emotion calculated by a predetermined arithmetic expression based on the degree of suppression and stimulation received from the emotion, the elapsed time, etc., and the current parameter value of the emotion as E [t] A coefficient representing the rate of changing the emotion according to the result etc. _e As follows:
[0047]
[Expression 1]

[0048]
Is used to calculate the parameter value E [t + 1] of the emotion in the next cycle.
[0049]
Then, the emotion / instinct model unit 31 updates the parameter value of the emotion so as to replace the calculation result with the current parameter value E [t] of the emotion. It should be noted that the emotion parameter value to be updated for each recognition result and each output action is determined in advance. For example, when a recognition result such as “struck” is given, the emotion of “anger” When the parameter value increases, the parameter value of the emotion of “joy” decreases, and when the recognition result such as “boiled” is given, the parameter value of the emotion of “joy” increases and “sadness” and “anger” The parameter value of each emotion is lowered.
[0050]
In the same manner, the emotion / instinct model unit 31 has five independent needs for “exercise greed”, “loving lust”, “appetite”, “curiosity” and “sleep greed” for each of these needs. A parameter value representing the strength of the desire is held. Then, the emotion / instinct model unit 31 sequentially updates the parameter values of each desire based on the recognition result from the state recognition unit 30, the elapsed time and the notification from the action determination unit 32, and the like.
[0051]
Specifically, the emotion / instinct model unit 31 calculates “exercise greed”, “loving greed”, and “curiosity” by a predetermined arithmetic expression based on the output behavior, elapsed time, recognition result, and the like of the pet robot 1. The change amount ΔI [k] of the desire, the current parameter value of the desire I [k], and the coefficient representing the sensitivity of the desire k _i As follows:
[0052]
[Expression 2]

[0053]
Is used to calculate the parameter value I [k + 1] of the desire in the next cycle, and the parameter value of the desire is updated so that the calculation result is replaced with the current parameter value I [k] of the desire. In this case, the parameter value of the desire to be changed with respect to the output action, the recognition result, or the like is determined in advance. For example, when there is a notification from the action determining unit 32 that some action has been expressed, The parameter value of “greed” decreases.
[0054]
Further, the emotion / instinct model unit 31 sets the remaining battery level for “appetite” to B based on the remaining battery level detection signal S2A (FIG. 2) given through the state recognition unit 30. _L As
[0055]
[Equation 3]

[0056]
Thus, the parameter value [k + 1] of “appetite” is calculated, and the parameter value of “appetite” is updated so that the calculation result is replaced with the parameter value I [k] of the current appetite.
[0057]
Further, the emotion / instinct model unit 41 increases / decreases the parameter value for every predetermined time with “sleep desire” as one synchronization per day.
[0058]
In the present embodiment, the parameter values of each emotion and each desire are regulated so as to vary in the range from 0 to 100, respectively, and the coefficient k _e , K _i The value of is also set individually for each emotion and each desire.
[0059]
(2-3) Configuration of the action determination unit 32
The behavior determining unit 32 has a plurality of behavior models in the memory 10A. And the action determination part 32 is based on the state recognition information S10 given from the state recognition part 30, the parameter values of each emotion and each desire in the emotion / instinct model part 31, the corresponding action model, the elapsed time, etc. The next action is determined, and the determination result is output to the emotion / instinct model unit 31, the action generation unit 33, and the growth control unit 34 as the action determination information S11. The details of the action determination unit 32 will be described later, and here, a method for determining the next action using the action model will be described briefly.
[0060]
In this pet robot 1, as a method for determining the next action, as shown in FIG. ₀ From the same or any other node NODE ₀ ~ NODE _n Each node NODE ₀ ~ NODE _n Arc ARC connecting the two ₀ ~ ARC _n Transition probability P set for each ₀ ~ P _n An algorithm called a stochastic automaton that is determined probabilistically based on the above is used.
[0061]
More specifically, the memory 10A stores, for each behavior model, each node NODE of the behavior model. ₀ ~ NODE _n Each state transition table 35 shown in FIG. 6 is stored.
[0062]
Here, in the state transition table 35, the node NODE ₀ ~ NODE _n Input events (recognition results of the state recognition unit 30) are listed in the “input event” row in priority order, and further conditions for the conditions correspond to the “data name” and “data range” rows. It is described in the column.
[0063]
Therefore, the node NODE defined in the state transition table 35 of FIG. ₁₀₀ Then, when the recognition result “ball detected (BALL)” is given, the “size (SIZE)” of the ball given together with the recognition result is “range from 0 to 100 (0, 1000)”. Or when the recognition result “OBSTACLE” is given, the “distance” to the obstacle given together with the recognition result is in the range of 0 to 1000 (0 , 1000) ”is self or another node NODE ₀ ~ NODE _n This is a condition for transitioning to.
[0064]
This node NODE ₁₀₀ Then, even when there is no input of the recognition result, “joy” or “SUPRISE” among the emotion and instinct model unit 31 parameter values of the emotions and instinct model unit 31 that the behavior determination unit 32 periodically refers to. ) ”Or“ Sadness (SDNESS) ”when the parameter value of the emotion is“ range from 50 to 100 (50, 100) ” ₀ ~ NODE _n It can transition to.
[0065]
Further, in the state transition table 35, the node NODE appears in the “transition destination node” column in the “transition probability to other node” column. ₀ ~ NODE _n Some nodes NODE that can transition from ₀ ~ NODE _n And the node NODE when all the conditions described in the “input event name”, “data value”, and “data range” lines are met. ₀ ~ NODE _n The node NODE in the column of “transition probability to other node” is the transition probability to ₀ ~ NODE _n The action and action output at this time are described in the “output action” line. Note that the sum of the transition probabilities of each row in the “transition probability to other node” column is 100 [%].
[0066]
Therefore, the node NODE in this example ₁₀₀ Then, for example, when a “ball is detected (BALL)” and a recognition result is given that the “size (SIZE)” of the ball is “range from 0 to 1000 (0, 1000)”, “ “NODE NODE” with a probability of 30% ₁₂₀ (Node 120) ", and the action or action of" ACTION 1 "is output at that time.
[0067]
Each behavior model is represented by a node NODE described as such a state transition table 35. ₀ ~ NODE _n Is formed to connect several.
[0068]
Thus, the action determination unit 32 receives the corresponding action stored in the memory 10 </ b> A when the state recognition information S <b> 10 is given from the state recognition unit 30 or when a certain time has elapsed since the last action was expressed. Corresponding node NODE in the model ₀ ~ NODE _n The state transition table 35 is used to probabilistically determine the next action or action (the action or action described in the “output action” line), and the determination result is used as the action decision information S11 for the emotion / instinct model unit 32. And output to the behavior generation unit 33 and the growth control unit 34.
[0069]
(2-4) Configuration of action generation unit 33
The behavior generation unit 33 corresponds to each “growth stage”, and the behavior and actions such as “walking state”, “motion”, and “sound” of “growth level” corresponding to the “growth stage”. 1 has various control parameters for causing it to appear in the memory 10A.
[0070]
Then, when the action determination information S11 is given from the action determination unit 32, the action generation unit 33, among these various control parameters stored in the memory 10A, “growth” designated in advance by the growth control unit 34 as described later. Necessary actuator 21 based on the control parameter of "stage" ₁ ~ 22 _n Is driven to a predetermined state, a sound based on the sound signal S3 is output from the speaker 20, or the LED 19 as an apparent “eye” is driven to blink in a predetermined pattern.
[0071]
In this way, the behavior generation unit 33 can cause the pet robot 1 to express the behavior determined by the behavior determination unit 32 at a level corresponding to the “growth stage” at that time.
[0072]
(2-5) Configuration of the growth control unit 34
As shown in FIG. 7A, the growth control unit 34 is the above-mentioned growth that should be used as a reference element when raising the “growth stage” among various states based on the state recognition information S10 given from the state recognition unit 30. A list of elements (hereinafter referred to as a first growth element list) 40A and a counter table (hereinafter referred to as a first growth element list) as shown in FIG. 41A) (referred to as a growth element counter table) in the memory 10A.
[0073]
Then, when the state recognition information S10 is given from the state recognition unit 30, the growth control unit 34 determines whether the state obtained based on the state recognition information S10 is a growth element based on the first growth element list 40A. If the state is a growth factor, the corresponding count value (experience value) in the first growth factor counter table 41A is incremented by one.
[0074]
Further, as shown in FIG. 8A, the growth control unit 34 is a reference element for raising the “growth stage” among the behaviors obtained based on the behavior decision information S11 given from the behavior decision unit 32 as described above. And a counter table (hereinafter referred to as FIG. 8B) for counting the cumulative frequency of these growth elements, respectively, and a list of growth elements to be described above (hereinafter referred to as a second growth element list) 40B. , This is called a second growth factor counter table) 41B in the memory 10A.
[0075]
Then, when the behavior determination information S11 is given from the behavior determination unit 32, the growth control unit 34 determines whether the behavior obtained based on the behavior determination information S11 is a growth factor based on the second growth factor list 40B. If the action is a growth factor, the corresponding count value (experience value) in the second growth factor counter table 41B is incremented by one.
[0076]
Further, when the growth control unit 34 increases the count value in the first or second growth factor counter table 41A, 41B as described above, the growth control unit 34 is different from the first and second growth factor counter tables 41A, 41B. The count value of a counter (hereinafter referred to as a total experience value counter) for determining whether or not to increase the “growth stage” separately prepared is incremented by 1, and thereafter the count value of the total experience value counter is It is determined whether or not a predetermined count value has been reached as an end condition for the “growth stage”.
[0077]
Then, when the count value of the total experience value counter reaches the preset count value as the current “growth stage” end condition, the growth control unit 34 changes the “character” of the pet robot 1 to the next “growth”. Which “personality” in the “stage” is determined based on the count values in the first and second growth factor counter tables 41A and 41B, and the determination result is used as the character designation information S20 and the action determination unit 32 and action generation To the unit 33. The growth control unit 34 provides personality designation information S20 indicating that the “personality” (Baby 1) of “childhood” should be selected to the behavior determination unit 32 and the behavior generation unit 33 at the initial stage.
[0078]
As a result, the behavior determining unit 32 selects the behavior model associated with the designated “personality” from the plurality of behavior models stored in the memory 10A based on the personality designation information S20. Uses the behavior model to determine the next behavior relative to the current behavior of the pet robot 1 as described above.
[0079]
The behavior generation unit 33 selects control parameters for “walking state”, “motion”, “sound”, and the like associated with the designated “growth stage” based on the personality designation information S20. Thereafter, each actuator 21 is used using this control parameter. ₁ ~ 21 _n Drive etc.
[0080]
In this way, the growth control unit 34 can control the “growth” of the pet robot 1.
[0081]
(3) Detailed configuration of the action determination unit 32
Next, a detailed configuration of the behavior determining unit 32 of the pet robot 1 will be described.
[0082]
In this pet robot 1, the behavior of the pet robot 1 is changed according to the contents as shown in FIG. 9 as “system system”, “falling / returning system”, “reaction system”, “obstacle avoidance system”, It is divided into several systems (hereinafter referred to as main classification systems) 50 to 55,... Such as “when high” and “basic behavior systems”, and each of these main classification systems 50 to 55,. One or a plurality of behavior models are prepared so that the behavior corresponding to the “growth stage” and “character” of the pet robot 1 can be expressed.
[0083]
Further, in this pet robot 1, the state of the pet robot 1 in the autonomous mode is played with a user and a ball, as shown in FIG. 10, such as a “bored state” in which there is no input from the user for a while. `` Naughty state '' that is when you are walking, `` Walking state '' where you are walking around looking for a ball etc., `` Guara state '' that is a state when you are lying around like bored It is divided into 6 basic states, “Talking / Teaching state” playing with the user in a sitting position, and “Good night state” such as when sleeping, and the behavior of these 6 basic states is expressed. The main classification system 55 of the “basic behavior system” is set to “boring system”, “tracking system”, and “walking system”, respectively, corresponding to the remaining five basic states except the “sleep state” that is not performed. The system is divided into five systems (hereinafter referred to as sub-classification systems) 55A to 55E of “sleeping system” and “interaction system”. A plurality of behavior models are prepared so that the corresponding behavior can be expressed.
[0084]
For example, when the above-described personality designation information S20 is given from the growth control unit 34, the behavior determination unit 32 sets the main classification systems 50 to 50 other than the “basic behavior system” in which two or more behavior models are prepared. For 54, one behavior model associated with the “personality” is selected in advance so that the behavior of the “personality” designated by the character designation information S20 can be expressed.
[0085]
For example, in the example of FIG. 9, since the designated “personality” was the “moist personality” (Child 1) (FIG. 3) in early childhood, the “falling recovery system” is restored when it falls. The behavior model of “calling without returning / fainting” that is easy to express the behavior of calling or fainting is selected, and the “obstacle avoidance system” is likely to exhibit the behavior that does not approach the obstacle. This shows that the behavior model “chicken” is selected. For “system system”, “reaction system”, and “system when emotion is slightly high”, only one behavior model is originally prepared, and therefore an action model of “Plane (normal)” is selected.
[0086]
At the same time, the action determination unit 32 also sets the “personality” specified by the character designation information S20 for each of the sub-classification systems 55A to 55E in the “basic behavior system” with respect to the main classification system 55 of the “basic behavior system”. The behavior models associated with the “personality” in advance are selected one by one so that the behavior “
[0087]
For example, in the example shown in FIG. 9, the “boring system” is “choro system”, the “tracking system” is “user system”, the “walking system” is “user search location moving system”, and the “sleeping system”. For "Susaya" and "Interaction", the action model of "Sunao" is selected.
[0088]
Then, the behavior determination unit 32 excludes the “basic behavior system” when the state recognition information S10 is given from the state recognition unit 30 or when a certain time has elapsed since the last action was expressed. The next behavior is determined using the behavior models of all corresponding main classification systems 50 to 54 among the main classification systems 50 to 54, respectively.
[0089]
Here, “all corresponding main classification systems 50 to 54” means main classification systems 50 to 54 that are associated in advance with the recognition result from the state recognition unit 30. Only the “system system” is associated with the recognition result from the state recognition unit 30 that “the internal temperature has become equal to or higher than the predetermined temperature” or “the remaining battery level is low”, and the recognition that “the system has fallen” Only the “falling recovery system” is associated with the result. In addition, “a system when the emotion is slightly high” and “a basic behavior system” are associated with a case where a certain time has passed since the last action was expressed. Accordingly, in this case, the subsequent actions are determined using the respective behavior models of “system when emotion is slightly high” and “basic behavior system”.
[0090]
In addition, the action determination unit 32 continues with the action model of one sub-classification system 55A to 55E associated with the recognition result from the state recognition unit 30 in advance for the “basic action system”. Determine action. For example, the “tracking system” is associated with the recognition result from the state recognition unit 30 that “the ball has been detected”. Therefore, in this case, the action model of the “tracking system” is used to determine the subsequent action. Is done.
[0091]
Then, the action determination unit 32 determines the next action using each action model of all the corresponding main classification systems 50 to 54 and one action model in the corresponding sub classification system 55A to 55E as described above. Then, one action selected by the main classification systems 50 to 55 having a predetermined high priority is selected from these determined actions, and the selected action is set as action determination information S11 as described above. The behavior generation unit 32 and the growth control unit 34 are notified. Incidentally, in this embodiment, the higher priority is set for the main classification systems 50 to 55 shown on the upper side in FIG.
[0092]
In this way, the behavior determination unit 32 can determine the subsequent behavior using the behavior model corresponding to the “character” of the pet robot 1 at that time.
[0093]
In this embodiment, even the pet robot 1 having the same “personality” in the same “growth stage” has difficulty and complexity of the output action according to the input history from the outside and the own action history. It is designed to make a change.
[0094]
That is, in the case of this pet robot 1, the action determining unit 32, as shown in FIG. 11, performs various tricks, a response to a loud sound (LOUD), a difficulty of movement (MOVE), a response to a ball (BALL), and a favorite Regarding some actions and actions such as reaction to color (FAV), the same action and action have several patterns with different levels of difficulty and complexity.
[0095]
When the growth control unit 34 raises the “growth stage” of the pet robot 1 as described above, the growth control unit 34 inside the first and second growth element counter tables 41A and 41B (FIGS. 7B and 8B). Based on each count value, the behavior determining unit 32 is notified of which pattern of behavior and motion should be used for the behavior and motion (hereinafter referred to as behavior difficulty designation notification). For example, FIG. 11 shows an example of the case where the “personality” is a “moist personality” (Child 1) (FIG. 3) in an early childhood. ”,“ Fainting ”is designated for a response to a loud sound,“ children ”is designated for the manner of contact with the ball, and“ children ”is designated for a hand.
[0096]
Further, the behavior determination unit 32 selects a designated behavior or action pattern based on the behavior difficulty designation notification, and the “basic behavior” shown in FIG. All behavior models that belong to the "system"("Chorosystem","Leisuresystem", ..., "Susaya system", "Vulgar system", ..., "Naughty system", "Suna system", "Independent system" , ..., "FAV domicile location micro system", "ball search location movement system", "user search location movement system", ..., "item system", "user system", ...) corresponding state transitions The corresponding output action column in Table 35 (FIG. 6) is changed to the corresponding output action.
[0097]
As a result, when the action or action is selected as the next action using the action model belonging to the “basic action system”, the action or action of the pattern designated by the growth control unit 34 is expressed.
[0098]
In this way, in this pet robot 1, even if it has the same “personality” in the same “growth stage”, the behavior and motion of different patterns (difficulty / complexity) are expressed for each behavior and motion. Thus, the pet robot 1 can have various personalities.
[0099]
(4) Operation and effect of the present embodiment
In the above configuration, in the pet robot 1, the behavior model for each basic state when the state of the pet robot 1 in the autonomous mode is divided into a plurality of basic states ("boring" of the "basic state system" in FIG. 9). System ”,“ tracking system ”,“ walking system ”,“ sleeping system ”, or“ interaction system ”), and the behavior corresponding to the recognition result of the state recognition unit 30 among these behavior models The behavior is determined by the behavior determination unit 32 using the model, and the actuator 21 is configured to express the determined behavior. ₁ ~ 21 _n Control etc.
[0100]
Accordingly, in this pet robot 1, the scale of the “basic state system” behavior model that has been one so far can be reduced to facilitate its handling, and the number of behavior models depends on how the basic state is divided. Therefore, the behavior of the pet robot 1 can be diversified.
[0101]
According to the above configuration, the behavior model for each basic state when the state of the pet robot 1 in the autonomous mode is divided into a plurality of basic states (“boring system”, “ Tracking system "," walking system "," sleeping system ", or" interaction system "), and among these behavior models, the behavior model according to the recognition result of the state recognition unit 30 is used. The action determining unit 32 determines the action, and the actuator 21 so as to express the determined action. ₁ ~ 21 _n Since the behavior model can be reduced and the handling thereof can be facilitated, and the number of behavior models can be increased depending on how the basic state is divided, the pet robot 1 Therefore, it is possible to realize a pet robot capable of diversifying the behavioral patterns of the robots and thus making it easy to change and diversify the behavioral patterns.
[0102]
(5) Other embodiments
In the above-described embodiment, the case where the present invention is applied to the quadruped walking pet robot configured as shown in FIG. 1 is described. However, the present invention is not limited to this, If it is an autonomous robot device that recognizes the external and / or internal state based on the output of the external sensor and / or the internal sensor and acts based on the recognition result, it can be widely applied to various other types of robot devices. can do.
[0103]
In the above-described embodiment, the basic states of the pet robot 1 in the autonomous mode are the “bored state”, the “naughty state”, the “walking state”, the “gutara state”, the “talking / telling state”, and the “ Although the case where the state is divided into six states of “sleeping state” has been described, the present invention is not limited to this, and may be divided into various other states, or may be divided into states other than 6. good.
[0104]
Further, in the above-described embodiment, the case where the memory 10A is applied as the storage means for storing various behavior models has been described. However, the present invention is not limited to this, and other storage means may be applied. May be.
[0105]
Furthermore, in the above-described embodiment, the action determination unit 32 as an action determination unit that determines an action using an action model according to an external and / or internal situation among various action models stored in the memory 10A; In accordance with the action generation unit 33 as a control unit that executes a predetermined control process so as to express the action determined by the action determination unit 32, the storage unit according to the external input history and / or the own action history Although the case where the growth controller 34 as the changing means for changing each of the stored behavior models is configured by the same controller 10 has been described, the present invention is not limited thereto, and these are configured separately. You may do it.
[0106]
Further, in the above-described embodiment, the case where the contents of each behavior model belonging to the “basic behavior system” shown in FIG. 9 is changed when raising the “growth stage” of the pet robot 1 has been described. The invention is not limited to this, and the timing may be a timing other than raising the “growth stage”. In this case, for example, the timing may be set by learning based on an external input history and / or own behavior history.
[0107]
【The invention's effect】
As described above, according to the present invention, At least one of external sensor and internal sensor Based on the output of At least one of the external and internal situations Recognize In addition, according to a counter that counts the cumulative degree of a plurality of growth factors related to a predetermined growth in each of the growth factors in at least one of the external and internal situations, While moving to the next growth stage, select a personality according to the cumulative degree of growth elements from a plurality of personalities with basic behavioral systems that are a series of behavior patterns, and select basic behavioral systems according to the cumulative degree of growth elements Select the behavior pattern with the set difficulty level from the basic behavior system with the selected personality for the growth control unit that sets the difficulty level for the behavior pattern belonging to and at least one of the external and internal situations A behavior control unit that controls the drive unit to act according to the behavior pattern selected by the behavior determination unit; By providing ,line Movement pattern Thus, a robot apparatus that can easily change and diversify behavior patterns can be realized.
[0108]
Also according to the invention, At least one of external sensor and internal sensor Based on the output of At least one of the external and internal situations Recognize In addition, a counter step for counting the cumulative degree of a plurality of growth factors relating to a predetermined growth in at least one of the external and internal situations for each growth factor, and depending on the total cumulative factor of the growth factors In addition to moving to the next growth stage, select a personality according to the cumulative degree of growth factors from a plurality of personalities with a basic behavior system that is a series of behavior patterns, and select basic behavior according to the cumulative degree of growth factors A growth control step for setting a difficulty level for an action pattern belonging to the system, and an action pattern having a set difficulty level from the basic action system for the selected character for at least one of the external and internal situations. An action determining step to select, and a drive control step to control the drive unit to act according to the action pattern selected by the action determining unit; By providing ,line Movement pattern Thus, a robot apparatus that can easily change and diversify behavior patterns can be realized.
[Brief description of the drawings]
FIG. 1 is a perspective view showing an external configuration of a pet robot according to an embodiment.
FIG. 2 is a block diagram showing an internal configuration of the pet robot according to the present embodiment.
FIG. 3 is a conceptual diagram for explaining “growth” of a pet robot.
FIG. 4 is a block diagram for explaining processing of a controller.
FIG. 5 is a conceptual diagram showing a stochastic automaton.
FIG. 6 is a conceptual diagram showing a state transition table.
FIG. 7 is a conceptual diagram for explaining a first growth factor list and a first growth factor counter table.
FIG. 8 is a conceptual diagram for explaining a second growth factor list and a second growth factor counter table.
FIG. 9 is a conceptual diagram for explaining an action determination unit.
FIG. 10 is a schematic diagram for explaining various basic states of the pet robot in the autonomous mode.
FIG. 11 is a schematic diagram for explaining various basic states of the pet robot in the autonomous mode.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 1 ... Pet robot, 10 ... Controller, 10A ... Memory, 30 ... State recognition part, 31 ... Emotion and instinct model part, 32 ... Action determination part, 33 ... Action generation part, 34 ... Growth Control unit, 50-55 ... main classification system, 55A-55E ... sub-classification system, S10 ... state recognition information, S11 ... action determination information, S20 ... personality designation information.

Claims (4)

Recognize at least one of the external and internal situations based on the output of at least one of the external and internal sensors, and determine a predetermined growth of at least one of the external and internal situations. A counter that counts the cumulative degree of a plurality of growth factors for each growth factor;
Depending on the total growth factor accumulation level, the system proceeds to the next growth stage, and selects a personality corresponding to the growth factor accumulation level from a plurality of personalities with basic behaviors that are a series of behavior patterns. A growth control unit that sets a difficulty level for an action pattern belonging to the basic action system according to a cumulative degree of the growth element;
An action determination unit that selects an action pattern having the set difficulty level from the basic action system in the selected personality for at least one of the external and internal situations;
And a drive control unit that controls the drive unit to act according to the behavior pattern selected by the behavior determination unit. The growth control unit
The robot apparatus according to claim 1 , wherein when moving to the next growth stage, the next character is selected according to the previous character. The drive unit is
The robot apparatus according to claim 2 , comprising an actuator that drives a hand or a foot in an animal type robot apparatus. Recognize at least one of the external and internal situations based on the output of at least one of the external and internal sensors, and determine a predetermined growth of at least one of the external and internal situations. A counter step for counting the cumulative degree of a plurality of growth factors for each growth factor;
Depending on the total growth factor accumulation level, the system proceeds to the next growth stage, and selects a personality corresponding to the growth factor accumulation level from a plurality of personalities with basic behaviors that are a series of behavior patterns. A growth control step for setting a difficulty level for the behavior pattern belonging to the basic behavior system according to a cumulative degree of the growth factor;
An action determining step of selecting an action pattern having the set difficulty level from the basic action system in the selected character for at least one of the external and internal situations;
A control method for a robot apparatus, comprising: a drive control step for controlling the drive unit so as to behave according to the behavior pattern selected by the behavior determination unit.

Images