Conference aamas: how to better pay attention to autonomous vehicles on fubgangers

Conference Aamas: How to better pay attention to autonomous vehicles on fubgangers

Fubgangers are greasy, at least from the perspective of cars: they run cross and transverse confused, look rare and show intended directional chants. On main hubs where the traffic streams of vehicles and passers-by are separated from each other and the Fubganger crosses the road only to particularly marked places, this is still a reasonable regulation.

In "Shared Spaces" against where all road users share a flat parts, without floor markers or traffic signs, the matter becomes harder. French researchers have now examined how autonomous vehicles can move safely in such a Kuddelmuddel.

Challenge for shared spaces

The concept of Shared Space has become quieter in recent years. It contradicts the focus of transport planning established for decades of transport planning on the needs of traffic and also requires a significant rethinking from the Fubgangern, as they have to pay attention to the traffic situation around the traffic situation.

In your contribution to the conference Aamas (Autonomous Agents and MultiAgent Systems), the four scientists around Manon Predhumeau (Universite Grenoble Alpes) refer to studies that have observed a reduction in accident numbers in Shared Spaces. For autonomous vehicles, however, such an environment is a special challenge due to the varied movement possibilities of the main road users who pursue them in the project Hianic (Human Inspired Autonomous Navigation in Crowds).

Social Force Model Combined with Decision Model

To model the movements of the Fubganger in a Shared Space, Predhumeau and their colleagues turn themselves on the Social Force Model. The motivations of the individual agents are considered as physical force: from the goal they seek, an attraction is based on obstacles on the way against the course. However, this model KONNE can only capture distant interactions with other road users, explain the researchers.

Therefore, it is therefore combined with a decision model that controls the reactions to immediately threatening collisions: Dodge when a vehicle comes from the front or back, stop or accelerate when it comes from the side. In addition, it applies to the group behavior to take into account. Many fubgangers did not move alone, but in groups, which is why another force must be introduced must, which considers such groups.

The comparison with real observed movements of Fubgangern showed that this hybrid model of their behavior simulated better than the pure social force model. So it could capture certain behaviors that escaped the Social Force Model: Race to cross the way from the vehicle’s nearby; Stop to stand by; stay together in the group, while the vehicle pass by.

Changes in the behavior of the Fubganger

For the course of course, the researchers want the parameters of the model (preferred running speed, safety chip, perceptual zones, etc.Refine and increase the diversitat of the individual agents and groups. They give concern that there are little observation data for traffic in Shared Spaces so far. In addition, the behavior of Fubgangern changed, the more she changed to autonomous vehicles, how the vehicles in turn will be adapted to the Fubganger. To what extent the model presented here can help an autonomous vehicle to safely navigate through unstructured crowds, should first be tested in simulations in a next step.

Like this post? Please share to your friends:
Leave a Reply

;-) :| :x :twisted: :smile: :shock: :sad: :roll: :razz: :oops: :o :mrgreen: :lol: :idea: :grin: :evil: :cry: :cool: :arrow: :???: :?: :!: