Fig 4 - uploaded by He Wang
Content may be subject to copyright.
(a) The neighborhood Ω(t) of a person is a sector within a circle (centered at this person with radius r col ) spanned by an angle ω from the current velocity vector (green arrow). (b) Each person has a view field (orange box) within which the environment repels a pedestrian. The view field is a square with dimension renv based on the current velocity vector (green arrow). The current velocity is along the diagonal of the orange box. (c) The environment is segmented into walkable (red) and unwalkable (blue) areas. Within the view field of the pedestrian in (b), the yellow pixels are the environment pixels that repel the pedestrian. ω, r col and renv are hyperparameters.

(a) The neighborhood Ω(t) of a person is a sector within a circle (centered at this person with radius r col ) spanned by an angle ω from the current velocity vector (green arrow). (b) Each person has a view field (orange box) within which the environment repels a pedestrian. The view field is a square with dimension renv based on the current velocity vector (green arrow). The current velocity is along the diagonal of the orange box. (c) The environment is segmented into walkable (red) and unwalkable (blue) areas. Within the view field of the pedestrian in (b), the yellow pixels are the environment pixels that repel the pedestrian. ω, r col and renv are hyperparameters.

Source publication
Preprint
Full-text available
Trajectory prediction has been widely pursued in many fields, and many model-based and model-free methods have been explored. The former include rule-based, geometric or optimization-based models, and the latter are mainly comprised of deep learning approaches. In this paper, we propose a new method combining both methodologies based on a new Neura...

Contexts in source publication

Context 1
... Repulsion. Pedestrians often steer to avoid potential collisions and maintain personal space when other people are in the immediate neighborhood (Fig. 4 a). Given an agent j in Ω t n of agent n and her state q t j , agent j repels agent n based on r nj = p t n − p t j ...
Context 2
... the environment is big, we assume the agent mainly focuses on her view field ( Fig. 4 b) within which the environment (Fig. 4 c) repels the pedestrian. We calculate p obs as the center of the pixels that are classified as obstacles in the view field of an agent. k env is shared among all obstacles. So far, we have introduced all the interpretable parameters θ = {τ, k nj , k env } in Equation ...
Context 3
... the environment is big, we assume the agent mainly focuses on her view field ( Fig. 4 b) within which the environment (Fig. 4 c) repels the pedestrian. We calculate p obs as the center of the pixels that are classified as obstacles in the view field of an agent. k env is shared among all obstacles. So far, we have introduced all the interpretable parameters θ = {τ, k nj , k env } in Equation ...
Context 4
... Repulsion. Pedestrians often steer to avoid potential collisions and maintain personal space when other people are in the immediate neighborhood (Fig. 4 a). Given an agent j in Ω t n of agent n and her state q t j , agent j repels agent n based on r nj = p t n − p t j ...
Context 5
... the environment is big, we assume the agent mainly focuses on her view field ( Fig. 4 b) within which the environment (Fig. 4 c) repels the pedestrian. We calculate p obs as the center of the pixels that are classified as obstacles in the view field of an agent. k env is shared among all obstacles. So far, we have introduced all the interpretable parameters θ = {τ, k nj , k env } in Equation ...
Context 6
... the environment is big, we assume the agent mainly focuses on her view field ( Fig. 4 b) within which the environment (Fig. 4 c) repels the pedestrian. We calculate p obs as the center of the pixels that are classified as obstacles in the view field of an agent. k env is shared among all obstacles. So far, we have introduced all the interpretable parameters θ = {τ, k nj , k env } in Equation ...