Abstract
Vision-based solution is a popular research for 6-DOF state estimation in recent years. All of these algorithms are based on the assumption that surrounding is static. Actually, in a highly dynamic environment, estimated state is mixed, including robot and dynamic objects. To this end, an approach for detecting dynamic objects based on the velocity of feature points is proposed in this paper. Our approach establishes the model of the velocity of feature points and separates dynamic objects motion from the mixed motion by comparing the velocity of IMU (inertial measurement unit) and feature points. Finally, the detection results of dynamic feature points on the left and right cameras’ front and rear frames are fused and verified to eliminate the influence of noise and mismatching. Experimental results prove that our approach can get more accurate state estimation, which shows that this method is effective in high dynamic environments.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call