A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
LEARNING SYSTEM, MOBILE BODY AUTONOMOUS OPERATION SYSTEM, LEARNING METHOD, AND MOBILE BODY AUTONOMOUS OPERATION METHOD
To generate a control model that permits a mobile body to perform autonomous operation on uneven ground.SOLUTION: A learning system that obtains a moving path of a mobile body travelling in a region including an uneven ground from a departure point to a destination and contents of operation control has a learning part that learns relation between environment data representing: three-dimensional point cloud data obtained by measuring the unevenness of the region; measurement data obtained by measuring a distance from a measurement position in the region to a surrounding object; and self-position data representing a position of the mobile body based on a reference point set in the region, and behavior data representing a wheel speed and wheel orientation of the mobile body moving along the surface of the region, and the learning part performs reinforcement learning of providing high reward when the mobile body moved from the departure point based on the behavior data and arrived at the destination while providing low reward when the mobile body overturned or crashed during its movement.SELECTED DRAWING: Figure 1
【課題】移動体が不整地を自律運転が可能な制御モデルを生成する。【解決手段】出発地から目的地までの間に不整地が含まれる領域を走行する移動体の移動経路と運転の制御内容とを求める学習システムであって、前記領域の凹凸が測定された３次元点群データと、前記領域における測定位置から周囲の物体までの距離が測定された測定データと、前記領域に設定される基準点をもとに移動体の位置を表す自己位置データとを表す環境データと、前記領域の面に沿って移動する移動体の車輪の速度と車輪の向きを表す行動データと、の関係を学習する学習部を有し、前記学習部は、前記移動体が前記行動データに基づいて前記出発地から移動し、前記目的地に到着した場合に高い報酬を、前記移動中に前記移動体が横転または衝突した場合に低い報酬を与える強化学習を行う。【選択図】図１
LEARNING SYSTEM, MOBILE BODY AUTONOMOUS OPERATION SYSTEM, LEARNING METHOD, AND MOBILE BODY AUTONOMOUS OPERATION METHOD
To generate a control model that permits a mobile body to perform autonomous operation on uneven ground.SOLUTION: A learning system that obtains a moving path of a mobile body travelling in a region including an uneven ground from a departure point to a destination and contents of operation control has a learning part that learns relation between environment data representing: three-dimensional point cloud data obtained by measuring the unevenness of the region; measurement data obtained by measuring a distance from a measurement position in the region to a surrounding object; and self-position data representing a position of the mobile body based on a reference point set in the region, and behavior data representing a wheel speed and wheel orientation of the mobile body moving along the surface of the region, and the learning part performs reinforcement learning of providing high reward when the mobile body moved from the departure point based on the behavior data and arrived at the destination while providing low reward when the mobile body overturned or crashed during its movement.SELECTED DRAWING: Figure 1
【課題】移動体が不整地を自律運転が可能な制御モデルを生成する。【解決手段】出発地から目的地までの間に不整地が含まれる領域を走行する移動体の移動経路と運転の制御内容とを求める学習システムであって、前記領域の凹凸が測定された３次元点群データと、前記領域における測定位置から周囲の物体までの距離が測定された測定データと、前記領域に設定される基準点をもとに移動体の位置を表す自己位置データとを表す環境データと、前記領域の面に沿って移動する移動体の車輪の速度と車輪の向きを表す行動データと、の関係を学習する学習部を有し、前記学習部は、前記移動体が前記行動データに基づいて前記出発地から移動し、前記目的地に到着した場合に高い報酬を、前記移動中に前記移動体が横転または衝突した場合に低い報酬を与える強化学習を行う。【選択図】図１
LEARNING SYSTEM, MOBILE BODY AUTONOMOUS OPERATION SYSTEM, LEARNING METHOD, AND MOBILE BODY AUTONOMOUS OPERATION METHOD
学習システム、移動体自律運転システム、学習方法、移動体自律運転方法
TARUYA AOI (author)
2024-11-07
Patent
Electronic Resource
Japanese
AUTONOMOUS MOBILE BODY CONTROL SYSTEM AND AUTONOMOUS MOBILE BODY CONTROL METHOD
| European Patent Office | 2024
MOBILE BODY SETTING DEVICE, MOBILE BODY, MOBILE BODY SETTING SYSTEM, METHOD, AND PROGRAM
| European Patent Office | 2024