Object detection on robosoccer environment using convolution neural network

Robots with autonomous capabilities depend on vision to detect and interact objects their environment. In the field of robotic research, one focus areas is robosoccer platform that being used implement test new ideas findings computer decision making. this article, an efficient real-time object detection algorithm employed in a simulation environment by deploying convolution neural network Kalman filter based tracking algorithms. This study's objective classify nao, ball, goalpost as well validate nao ball without human intervention from initial frame last frame. comparison existing methods, proposed method robust fast identifying three classes namely speed 1.67 FPS mAP 95.18%. By implementing approach, soccer playing robots can make appropriate decisions during game play.