People Trajectory Forecasting and Collision Avoidance in First-Person Viewpoint
Guan-Yu Lai, Kuan-Hung Chen, Bau‐Jy Liang · 2018
we propose a new collision avoidance system for first-person viewpoint, to show trajectory of people and to predict the future location of them. Then, we can determine the predicted location to avoid collision. We use deep learning to detect pedestrians and plot out coordinates of the trajectory. We predict future location of the target according to the law of inertia. In the first-person screen, this system can show whether possible collision occurs. Experimental results show that our method is feasible and outperforms state-of-the-art.