基于神经网络集成、多传感器融合的机器人对障碍物的识别

Mobile Robot Avoid Barrier Based on Neural Network Integration and Multi-sensor Fusion

基于神经网络和神经网络集成理论提出了一种多传感器信息的数据融合结构,并将其用于机器人的障碍物的识别,提高了系统的识别效率,增强了系统的可靠性。通过分别搭建识别各种障碍物的子网络,以并行集成的方式把各个个体网络组合起来,可以获得一个一个高性能的识别系统。在HEBUT-Ⅰ型移动机器人上进行了验证,取得了很好的识别效果,为机器人的正确导航奠定了基础。

Since more information can be obtained and the ability of target identification can be enhanced with multi-sensor information fusion technology, thus the limitation of a single sensor system can be avoided. The paper develops a kind of structure of multi-sensor information date fusion based on Neural Network and Neural Network integration. It can improve the robustness and reliability.The simulation with the robotic multi-sensor barrier identification is also presented and discussed.