Replace the obstacle avoidance of steering behavior with RVO, is it possible?
0
Is it possible to combine steering behavior and Reciprocal Velocity Obstacle
Started by pandaa, Jul 17 2012 09:28 PM
5 replies to this topic
Sponsor:
#2 Members - Reputation: 169
Posted 20 July 2012 - 03:29 AM
RVO is A steering behaviour, every steering behaviour can be combined to obtain the result you want. When no obstacles your point move according to its behaviour, when you find obstacle the point avoid the obstacle using RVO
Why not using reputation system? thumbs up thanks 
#3 Members - Reputation: 150
Posted 20 July 2012 - 06:50 AM
I mean steering behavior is http://www.red3d.com/cwr/steer/RVO is A steering behaviour, every steering behaviour can be combined to obtain the result you want. When no obstacles your point move according to its behaviour, when you find obstacle the point avoid the obstacle using RVO
#5 Members - Reputation: 1257
Posted 23 July 2012 - 04:10 AM
Not really. Velocity obstacle algorithms use velocities as an input to compute regions in the plane; control of your entity in order to avoid those regions is a conceptually separate step and it is achieved by applying forces, exactly like in the case of traditional steering behaviours and any other type of physically decent dynamic simulation.steering behavior is force-based, RVO is velocity-based
You might be confused by the fact that many steering behaviours compute forces rather directly by adding up components instead of explicitly computing the appropriate acceleration for a desired velocity and location like in typical VO setups, but they have an equally advisory value; for example, these forces should be realistically clipped to a certain maximum force, further reduced to avoid exceeding velocity limits, and so on.
You can combine RVOs and steering behaviours in a number of ways, most likely by giving priority to RVO-based collision avoidance: for example, you might compute the force your steering behaviours would like to apply and correct it by braking and by swerving towards the "best" side of the RVO if it would cause a collision.
Produci, consuma, crepa
#6 Members - Reputation: 178
Posted 23 July 2012 - 11:53 PM
Not really. Velocity obstacle algorithms use velocities as an input to compute regions in the plane; control of your entity in order to avoid those regions is a conceptually separate step and it is achieved by applying forces, exactly like in the case of traditional steering behaviours and any other type of physically decent dynamic simulation.
steering behavior is force-based, RVO is velocity-based
You might be confused by the fact that many steering behaviours compute forces rather directly by adding up components instead of explicitly computing the appropriate acceleration for a desired velocity and location like in typical VO setups, but they have an equally advisory value; for example, these forces should be realistically clipped to a certain maximum force, further reduced to avoid exceeding velocity limits, and so on.
You can combine RVOs and steering behaviours in a number of ways, most likely by giving priority to RVO-based collision avoidance: for example, you might compute the force your steering behaviours would like to apply and correct it by braking and by swerving towards the "best" side of the RVO if it would cause a collision.
thanks, I'll have a try.
