Trajectory prediction considering the behavior of pedestrians intersecting with vehicles

AbstractIn this paper, we propose a trajectory prediction method that takes into account pedestrian behavior. To realize safe automated driving in urban areas, it is necessary to predict the future movements of road users. Pedestrians entering the vehicle's direction are a target of interest, but they are difficult to predict because they change their behavior through significant interactions with the vehicle. In this study, we first predict whether pedestrians will yield the way to a vehicle. Next, the predictions are then used to predict the future trajectories of all road users in the scene. The proposed model consists of two neural network structures: the Yielding judgment module to predict pedestrian behavior and the Trajectory prediction module. The yielding judgment module provides intuitive and easily understandable indicators, which also helps to increase the interpretability of the overall model. We evaluated the usefulness of the proposed model using a publicly available dataset. The proposed method was found to reduce the average displacement error by 2.79% and the final displacement error by 2.70% compared to the case where the target pedestrian's behavior is not considered.Keywords: Trajectory predictionpedestrian-vehicle interactionrecurrent neural networkattention mechanism Disclosure statementNo potential conflict of interest was reported by the author(s).Additional informationNotes on contributorsSoichiro TannoSoichiro Tanno received his BE degree from the Department of Mechanical and Aerospace Engineering, in Tohoku University, Japan in 2022. He is currently a Master student at the Smart Robots Design Laboratory of Tohoku University. He is working on topics related to Pedestrian Prediction for Safe and Comfortable Coexistence of Humans and Robots.Yusuke TamuraYusuke Tamura received his BE, ME, and PhD degrees in Engineering in 2003, 2005, and 2008, respectively, from the University of Tokyo. From 2006 to 2008, he was a Research Fellow of the Japan Society for the Promotion of Science. He worked as a Project Researcher at the University of Tokyo, Japan from 2008 to 2012 and as an Assistant Professor at Chuo University, Japan from 2012 to 2015. From 2015 to 2020, he was a Project Associate Professor at the University of Tokyo. He is currently an Associate Professor at the Department of Robotics, Tohoku University, Japan. His research interests include human-robot interaction and mobile robotics. He is a member of IEEE, JSME, and RSJ.Yasuhisa HirataYasuhisa Hirata is a Professor in the Department of Robotics at Tohoku University, Japan, and a Project Manager of the Moonshot R&D program in Japan. He received his B.E., M.E., and Ph.D. degrees in mechanical engineering from Tohoku University in 1998, 2000, and 2004, respectively. He formerly worked as a research associate and an associate professor at Tohoku University. He was also a visiting researcher at The Universite de Versailles Saint-Quentin-enYvelines, France in 2006 and 2012. He served as an AdCom member and vice president of TAB in IEEE RAS. For more than 20 years, he has been doing research on the control of multiple mobile robots in coordination, human-robot cooperation systems, assistive robots, haptics, industrial robots, etc. He has over 200 technical publications in the area of robotics. He received the Best Paper Awards in Advanced Robotics, JSME Journal, RSJ Journal, Fanuc FA Foundation, ROBIO 2004, ICMA 2020, etc.