Behavior Control of Apelike Robot Based on Multi-Sensor Integration

去年，我们开发了仿生机器人和视觉控制系统。本文研究了基于传感器图像提取的多传感器数据集成技术。我们结合多传感器数据，生成完整的传感器图像，并与类猿机器人进行了真实行为的实验。研究结果如下：1.传感器图像生成电路为规范多传感器信息处理，研制了将传感器数据融合到NTSC视频图像上的硬件电路。多传感器数据融合软件的开发利用快速跟踪模块对传感器图像进行在线处理。首先，为了通过检测机器人的姿态和运动来获得平衡信息，首先从图像处理中获得a)倒下开关，b)姿态传感器，c)姿态定向。其次，研制了分布于全身的触觉传感器。最后，开发了一个可以处理任意传感器信息的顶层软件环境。基于多传感器信息的实验行为通过研究机器人基本运动时的传感器图像模式，实现了基于传感器图像时间序列变化的类人猿运动控制实验。通过本文的研究，展示了传感器图像方法的优势。

Last year we developed aplike robot and vision control system. The current year, multiple sensor data integration based on the sensor image apporach are studied. We combined multiple-sensor data, generated integrated sensor image, and did experiment of the real behavior with and apelike robot. The results are as follows,1. Sensor Image Generation CircuitIn order to standardize multiple sensor information processing, we developed hardware circuit that combined sensor data ionto the NTSC video image.2. Development of Multiple-Sensor Data Integration SoftwareSensor image are processed on-line using fast tracking module. At first, in order to obtain balancing information by detecting robot posture and movement, a) falling down switch, b) posture sensor, c) posture derection from image processing are atacched. Next tactile sensor that distributed all over the body are developed. Finally, an environment in which top-level software can treat any sensor information are developed.3. Experimentla Behavior based on the Multiple-Sensor ImformationExperiments of apelike motion control based on sensor image changes in time sequences are succeeded by investigating sensor image pattern while basic motion of the robot. Through out this study the advantages of sensor image approach are shown.