The use of norms violations to model agents behavioral variety Benot - - PowerPoint PPT Presentation

▶
the use of norms violations to model agents behavioral
SMART_READER_LITE
LIVE PREVIEW

The use of norms violations to model agents behavioral variety Benot - - PowerPoint PPT Presentation

Jun 13, 2023 238 likes •427 views

The use of norms violations to model agents behavioral variety Benot Lacroix 1,2 , Philippe Mathieu 2 and Andras Kemeny 1 COIN@AAMAS 2008 2 LIFL, University of Sciences and Technologies of Lille 1 Technical Center for Simulation, Renault

slide-1
SLIDE 1

COIN@AAMAS 2008 May 12th, 2008

1 Technical Center for Simulation, Renault 2 LIFL, University of Sciences and Technologies of Lille

The use of norms’ violations to model agents’ behavioral variety

Benoît Lacroix 1,2, Philippe Mathieu 2 and Andras Kemeny 1

slide-2
SLIDE 2

2

Benoît Lacroix COIN@AAMAS 2008 May 12th, 2008

Motivation

  • LIFL / Renault collaboration
  • Traffic simulation in driving simulators
  • Needs

Improve the behavioral realism of autonomous vehicles Obtain consistent and various behaviors

  • Idea

Driving psychologists classify drivers depending on their behavior (Saad,

1992)

The Highway Code is a set of norms Drivers do not strictly follow the Highway Code

= model behavioral variety as violations

  • f the norms
slide-3
SLIDE 3

3

Benoît Lacroix COIN@AAMAS 2008 May 12th, 2008

Session context

Determine fulfillment and violations of expectations without recourse to future information

Not possible to see future on real-time applications with human in the loop Strong interest in simulation to analyse how humans behave and adapt agents’ behavior

Protect agents from potentially harmful interactions

Consistency with characteristics and capabilities of the agents Manage the consistency of the agent itself

slide-4
SLIDE 4

4

Benoît Lacroix COIN@AAMAS 2008 May 12th, 2008

Plan

General context and proposed approach Related work Institutional environment Application to traffic simulation Conclusion and future work

slide-5
SLIDE 5

5

Benoît Lacroix COIN@AAMAS 2008 May 12th, 2008

General context and approach

General context: simulation of spatially situated agents

Microscopic simulations Behaviors’ emergence through interactions A need of variety and consistency for realism

Our approach

Use norms to generate the characteristics of the agents No regulation of their behavior Based on norms’ definitions and norms’ violations

slide-6
SLIDE 6

6

Benoît Lacroix COIN@AAMAS 2008 May 12th, 2008

Related work

Institutions and norms (Noriega, 1997; Esteva et al., 2001)

Regulation and management of agents’ social interactions Authority and control instances

Improvement of traffic control strategies (Bou et al., 2007)

Autonomic capabilities of institution Application to traffic

Non-normative behaviors in traffic simulation (Doniec et

al., 2006)

Introduce formal rules of the road Agents allowed to break them

slide-7
SLIDE 7

7

Benoît Lacroix COIN@AAMAS 2008 May 12th, 2008

Institutional environment

  • Replication of properties allows specifying elements
  • No control instance (handled by the execution model)
slide-8
SLIDE 8

8

Benoît Lacroix COIN@AAMAS 2008 May 12th, 2008

Behavior variety

  • Through norms’ definitions

While instantiating behaviors

  • By violating the norm

Parameters generation outside the definition domains Consistency no more guarantied Able to quantify the deviation

  • Choice between allowing or not to violate norms
  • Conflicting norms

Norms does not rule the whole environment Handled by the underlying execution model

slide-9
SLIDE 9

9

Benoît Lacroix COIN@AAMAS 2008 May 12th, 2008

Application: driving simulators

  • Ergonomics, embedded systems, design, headlights…
  • Agents’ decision model: perception – decision (finite state

automata) – action (vehicle dynamic model)

slide-10
SLIDE 10

10

Benoît Lacroix COIN@AAMAS 2008 May 12th, 2008

Traffic simulation

Drivers behavior

Individual differences (Dewar, 2002) Psychological factors Formal rules, informal rules, design of the road and

  • ther drivers behavior (Björklung, 2005)

Road system

Highway code: set of norms Country, city, local rules

slide-11
SLIDE 11

11

Benoît Lacroix COIN@AAMAS 2008 May 12th, 2008

Current implementation

Choices

Norms does not manage the whole traffic logic Uses the existing traffic model (finite state automata) Norms based on driver parameters

Current norms

Define normal, aggressive, cautious and elderly drivers Easily extensible by driving psychologists

slide-12
SLIDE 12

12

Benoît Lacroix COIN@AAMAS 2008 May 12th, 2008

Example

slide-13
SLIDE 13

13

Benoît Lacroix COIN@AAMAS 2008 May 12th, 2008

slide-14
SLIDE 14

14

Benoît Lacroix COIN@AAMAS 2008 May 12th, 2008

slide-15
SLIDE 15

15

Benoît Lacroix COIN@AAMAS 2008 May 12th, 2008

Conclusion and future work

Conclusion

Able to guaranty the consistency of the parameters Realistic behaviors without hard-coded scenarisation

Future work

Macroscopic validation with real data Introduction of atomic norms defining behaviors (speeding, relative to environment (city, highway)…) Study the effect of norms’ violation on the global system, as well as norms proportion influence

slide-16
SLIDE 16

16

Benoît Lacroix COIN@AAMAS 2008 May 12th, 2008

Thank you for your attention

Contact: benoit.lacroix@renault.com

slide-17
SLIDE 17

17

Benoît Lacroix COIN@AAMAS 2008 May 12th, 2008

Questions?

As an agent is just a piece of software, how is it possible for it to violate a norm? It only does what I defined ;-) Isn’t « outside of norms » also a norm? Norms are defined behaviors, you have a name for them or not, but you are always in one of them! How to qualify norms violations?