Multiple human trajectory prediction and cooperative navigation modeling in crowded scenes
- PDF / 1,835,656 Bytes
- 15 Pages / 595.276 x 790.866 pts Page_size
- 34 Downloads / 215 Views
ORIGINAL RESEARCH PAPER
Multiple human trajectory prediction and cooperative navigation modeling in crowded scenes Akif Hacinecipoglu1
· E. Ilhan Konukseven2 · A. Bugra Koku2
Received: 22 August 2019 / Accepted: 13 July 2020 © Springer-Verlag GmbH Germany, part of Springer Nature 2020
Abstract As mobile robots start operating in environments crowded with humans, human-aware navigation is required to make these robots navigate safely, efficiently and in socially compliant manner. People navigate in an interactive and cooperative fashion so that, they are able to find their path to a destination even if there is no clear route leading to it. There are significant efforts to solve this problem for mobile robots; however, they are not scalable to high human density and learning based approaches depend heavily on the context and configuration of the set they are trained with. We develop a method which infers initial trajectories from Gaussian processes and updates these trajectories jointly for all agents using a cost based interaction approach. We condition Gaussian processes online with the best hypothesis at each step of prediction horizon. The method is tested on a common public dataset and it is shown that it outperforms two state-of-the-art approaches in terms of human-likeness of predicted trajectories. Keywords Human navigation · Trajectory prediction · Mobile robots · Path planning
1 Introduction In dynamic environments, especially crowded with people, it is a challenging task for a robot to navigate to a goal position. A robot should be able to predict the trajectories of other agents to navigate safely, efficiently and in socially complaint manner. When this prediction is carried out to an extent, uncertainty grows at each prediction step. This uncertainty explosion is one of the reasons for preventing robots from generating reasonable trajectories for navigating among humans. In this case, robot fails to plan its trajectory to its destination and halts motion or carries out inefficient movements. This is called as the freezing robot problem in the literature [1]. Although uncertainty explosion is the primary reason behind the freezing robot problem, it is shown that even in perfect knowledge of individual trajectories (or with insignificant uncertainty), problem still occurs in certain crowd configurations. Studies aimed to decrease uncertainty in predictions by developing better human motion models [2– 4]; however, it is shown that if the crowd is dense enough,
B
Akif Hacinecipoglu [email protected]
1
OSTIM Technical University, Ankara, Turkey
2
Middle East Technical University, Ankara, Turkey
the freezing robot problem will always occur [1]. The reason behind this is the lack of human cooperation in path planning models. For example, when people walk shoulder to shoulder any independent planner will assume the path is blocked and will try to navigate around the crowd. However, in human navigation, even while walking shoulder to shoulder, humans cooperate and make room for others [5–7]. Lack of
Data Loading...
