Discrepancies between SteerSuite RVO and original RVO model

[John Charlton]

There appears to be some differences between the RVO model implemented in SteerSuite, compared to the original model (source from GitHub: https://github.com/snape/RVO2.git). A video shows differences between the models https://youtu.be/HO6n4baCCns. I was wondering what could cause this. 

Thank,

John

[Glen B]

SteetSuite has a more complex path planning library than RVO may have had. Also, we technically use the ORCA/RVO2 algorithm not RVO. It might also be related to the timestep the simulation uses. We often use 20hz, not sure what RVO uses.

Take care.

--
You received this message because you are subscribed to the Google Groups "SteerSuite" group.
To unsubscribe from this group and stop receiving emails from it, send an email to steersuite+...@googlegroups.com.
To post to this group, send email to steer...@googlegroups.com.
Visit this group at https://groups.google.com/group/steersuite.
For more options, visit https://groups.google.com/d/optout.

--

Glen Berseth

http://www.fracturedplane.com

[John Charlton]

As far as I am aware, I have tested the SteerSuite RVO module with the same parameters (defined in RVO2D_Parameters.h)  and in the SNAPE/RVO2 repository, including time step. I have also disabled the more complex path planning by setting dont_plan=true in rvo2AI/src/RVO2DAIModule.cpp line 77. I would have thought that this would mean the behaviors between the two models should be the same, yet they are not.

[Glen B]

There are many reasons the simulations will be exactly the same. Libraries could use different units/distances, numerical error, different spatial databases, etc. My guess here is that the agents are processed in different orders. Each agents computes its steer decision one at a time. The order can greatly influence the outcome.

On Thu, Aug 23, 2018, 7:16 AM John Charlton, <j.a.ch...@sheffield.ac.uk> wrote:

As far as I am aware, I have tested the SteerSuite RVO module with the same parameters (defined in RVO2D_Parameters.h)  and in the SNAPE/RVO2 repository, including time step. I have also disabled the more complex path planning by setting dont_plan=true in rvo2AI/src/RVO2DAIModule.cpp line 77. I would have thought that this would mean the behaviors between the two models should be the same, yet they are not.

--

You received this message because you are subscribed to the Google Groups "SteerSuite" group.
To unsubscribe from this group and stop receiving emails from it, send an email to steersuite+...@googlegroups.com.
To post to this group, send email to steer...@googlegroups.com.
Visit this group at https://groups.google.com/group/steersuite.
For more options, visit https://groups.google.com/d/optout.