Fig 5 - uploaded by He Wang
Content may be subject to copyright.
Download
View publication
Red dots are observed, green dots are our prediction and black dots are the ground-truth. Blue dots are pedestrians. F goal , F col and Fenv are shown as yellow, light blue and black arrows for a person. The orange areas are the view field for avoiding collisions with other people (left) and the environment (middle). They provide plausible explanations of individual behaviors such as steering. Left and middle show the major influence of different forces. Right shows motion randomness captured by our model.

Red dots are observed, green dots are our prediction and black dots are the ground-truth. Blue dots are pedestrians. F goal , F col and Fenv are shown as yellow, light blue and black arrows for a person. The orange areas are the view field for avoiding collisions with other people (left) and the environment (middle). They provide plausible explanations of individual behaviors such as steering. Left and middle show the major influence of different forces. Right shows motion randomness captured by our model.

Source publication
Fig. 2. Left: Goal-Network and Right: Collision-Network. The numbers in...
Fig. 4. (a) The neighborhood Ω(t) of a person is a sector within a...
Fig. 5. Red dots are observed, green dots are our prediction and black...
Fig. 6. Red, green and cyan dots are observations, prediction and...
+6 Fig. 7. The collision rate and the number of collisions against the...
Human Trajectory Prediction via Neural Social Physics
Preprint
Full-text available
  • Jul 2022
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...
Cite
Download full-text

Contexts in source publication

Context 1
... analysis [72]. Figure 5 Left shows that a person, instead of directly walking towards the goal, steered upwards (the green trajectory in the orange Table 4. Collision rates of the generalization experiments on ZARA2 (Z) and coupa0 (C). NSP-SFM shows strong generalizability in unseen high density scenarios. ...
Context 2
... could be explained by the strong repulsive force (the light blue arrow) which is generated by the potential collisions with the agents in front of this person, in line with existing studies [41]. Similar explanations can be made in Figure 5 Middle, where all three forces are present. F env (the black arrow) is the most prominent, as expected, as the person is very close to the car. ...
Context 3
... repulsive force (light blue arrow) also plays a role due to the person in front of the agent (the blue dot in the orange area). Figure 5 Right shows an example where motion randomness is captured by NSP. In this example, there was no other pedestrian and the person was not close to any obstacle. ...
Context 4
... analysis [72]. Figure 5 Left shows that a person, instead of directly walking towards the goal, steered upwards (the green trajectory in the orange Table 4. Collision rates of the generalization experiments on ZARA2 (Z) and coupa0 (C). NSP-SFM shows strong generalizability in unseen high density scenarios. ...
Context 5
... could be explained by the strong repulsive force (the light blue arrow) which is generated by the potential collisions with the agents in front of this person, in line with existing studies [41]. Similar explanations can be made in Figure 5 Middle, where all three forces are present. F env (the black arrow) is the most prominent, as expected, as the person is very close to the car. ...
Context 6
... repulsive force (light blue arrow) also plays a role due to the person in front of the agent (the blue dot in the orange area). Figure 5 Right shows an example where motion randomness is captured by NSP. In this example, there was no other pedestrian and the person was not close to any obstacle. ...