Four Different Model of Crowd Control - Lab Report Example

Note: Just send me a message if you need anything. Thanks CROWD CONTROL MODELS Literature Review 1. Decision Support for Crowd Control – Genetic Algorithms with Simulation to Learn Control Strategies Important Details Model characteristic Description Comments Crowd modelling type Macroscopic/decision support Crowd movement /interaction Crowd AI Crowd simulation technique Stochastic agent based Real-time application capability No Not intended for real-time application. Model simulation result and control strategies are intended for barrier design and positioning only. Crowd movement is not given much emphasis but rather focus on destructive force or effects to barriers and structures Contribution to knowledge & technology Learning and decision-making algorithm Proof-of-concept Proof-of-concept is introduced to enhance existing stochastic models. New agent learning and decision-making techniques Usefulness for future study/experiments - Barrier strategic positioning - Barrier design and strength - Control strategies and proof-of-concept Can be useful for actual design and installation of barriers and other crowd control system Developed by Shubert & Suzic (2007) for IEEE, this model intends to provide a decision support system for crowd control specifically for riot situation. The strategy is to deploy barriers in position with respective strengths to effectively control the rioting crowd. The simulated situation can be use as reference in the prediction and development of actual control strategies particularly with the police. The model derived its control strategies from pre-stored example of situations using genetic algorithm while stochastic agent-based simulation is used to validate the strategies. In other words, these strategies correspond to every possible strength or force that may be encountered by a predetermined barrier position. On the other hand, the genetic algorithm provides the characteristics of all individuals that may be involved in the riot which will be the basis of barrier position and manning. The model was originally developed with stochastic agent models in mind but added a “proof-of-concept” model to enhance embedded simulations. The inputs in this simulation are strategies and positions of real-world agents while the predicted effects are outputs. The model considered the impact of hostilities and the destruction it would to installed barriers and nearby structures. Validation is done through rating or scoring each strategy determined by the intensity of the above mentioned impact. The decision support and control strategies are shown below. Figure 1- Decision support and control strategies The decision support is intended to determine the optimal control strategy by calculating the agent starting position , axis, distance to travel, variance of its position, and the situation in that particular position in the fuzzified position map. The genetic algorithm is responsible for evolving the behaviour of the agents’ population through recursive incremental modifications. These include learning and decision-making algorithms. For example, the algorithm with select two agents population randomly and generate next generation offspring from them. An agent in the simulation is considered anything, individual or group that has sensors and effectors representing the real-world agents such as rioters, peaceful protesters, and the police as shown below. Figure 3- Stochastic agent-based simulation 2. Behavioural Animation Model Important Details Model characteristic Description Comments Crowd modelling type Macroscopic/behavioural Crowd movement /interaction Crowd AI Crowd simulation technique Real-time agent based Real-time application capability Yes Real-time agent motion and crowd flow using 3D models and scenario. Real-life characters with respective behavioural mechanism, bio-mechanical movements and gestures. Contribution to knowledge & technology Multi-model approach and introduction of GASP & HPTS Usefulness for future study/experiments -Generation of real-life characters - Database of agent states & perception - Algorithms for bio-mechanical movements and gesture - Real-life 3D characters can be used for other experiments using 3D modelling software - Data from databases can be use for future study such as states, perception, and algorithms The idea behind Donikan (2000) behavioural animation model is to simulate autonomous beings like people and organisms. Similar to real-time agent-based simulation it is dependent on behavioural characteristics and forces in the environment – decision, action, communication, perception, and so on. However, this model is more complex since incorporates various models – environment, mechanical, motion control, sensor models, geometric models, and various scenarios. The main simulation environment is GASP or General Animation and Simulation Platform while HPTS or Hierarchical Parallel Transition Systems is used to model the behavioural patterns. The state in which the agent is in is represented by parameters RANDOM, DELAY, WEIGHT, and TIMEGATE which is actually the transitional behaviour of the agent reacting to forces. The model generate realistic geometrical model through a 3D modelling system which walk through is possible in real time as shown by the model’s application and Virtual Museum below. The principle used in this model is to have real-life characters (attendant, thief, sculptures, photographer, mother and son, and so on) interacting with an agent (visitor) so life can observed as well as behaviour modification of characters as they are introduced to pre-emption mechanism , dynamics, and parameters mentioned earlier. The simulation is main influence by Sense-Control-Action or SCA loops with accompanying planners. The loops determine the reflexive behaviour through a network of nodes interconnecting sensors and effectors. Through a database of states and perceptions, the agent selects its perception, plan, and action. The motion as shown in the figures above is humanoid motion control which consists of motion algorithms mimicking bio-mechanical movements and gestures. 3. SCADA System for Crowd Control Important Details Model characteristic Description Comments Crowd modelling type Macroscopic/closed loop Crowd movement /interaction Density count/space signature Crowd simulation technique Real-time monitoring/surveillance Real-time application capability Yes CCD Camera for real-time crowd monitoring, surveillance, and control. Real-time density count using image background frame differencing and space signatures through agents form, dimension, and luminosity. Contribution to knowledge & technology Background frame differencing/ image fine and coarse pattern density modelling Usefulness for future study/experiments Pattern matching Frame differencing Space signature -Data can be use to match patterns in other scenario - Other study can use or improve (remove errors caused by objects with similar intensity) frame differencing technique and space signature which is somewhat innovative. The SCADA system is generally intended for crowd monitoring, surveillance and control. It uses sensors, algorithms for automatic crowd estimation, and pedestrian control. The idea behind this system is managing and controlling the crowd using a closed loop as shown below. The sensor and DAS in the above closed loop will as counters for pedestrians while FLC or the fuzzy logic system will provide the required diameter to control the flow of agents in the designated area. To detect the density of the crowd, this model uses a crowd monitoring sensor which is actually a CCD camera focus on the area being monitored. The image is then recorded and process using background frame differencing. This means separating moving objects from their background. The idea is to count the number of foreground picture elements that represent the number of people in the monitored area. Since there is a linear relationship between the number of picture and edge elements to the actual quantity of people and fine and coarse patterns often represent the high and low density of the crowded area respectively, the technique can estimate the actual density or number of people in the crowd. Similarly, each pedestrians or people moving to and from the monitored area can identify by form, dimensions, and luminosity or “space signature” in general. Another technique to identify them is template matching (a previously taken image with same dimension, form, and so on) that can be compared with the newly recorded images. 4. Feedback for Crowd Control Important Details Model characteristic Description Comments Crowd modelling type Macroscopic/Crowd dynamics Crowd movement /interaction Agent-based with signals Crowd simulation technique Atkeson’s Game-Theoretic with feedback Real-time application capability No Only intended for behavioural overtime graphing of crowd responses to signal. Only useful for planning and development of controlling forces strategy. No real-time crowd monitoring and surveillance features or 3D modelling applied. Contribution to knowledge & technology Feedback mechanism Signals or strength/penalty thresholds for balancing crowd and controlling forces equilibrium Usefulness for future study/experiments Feedback approach Signalling technique Feedback and signalling approach can be very useful for studies interested in maintain equilibrium than using force to control a crowd . Developed through the joint effort of Taylor et al. (2007), this crowd control model is intended to control a large crowd by simulating crowd dynamics in confrontation situation. The model is dependent on two known crowd modelling paradigms – System Dynamics and Agent-Based Models. The first model is adapted from game-theoretic approach which is using the Theory of Rational Choice. The idea comes from the possibility that independent decisions of agents in rioting crowd will be influenced by the potential harm that will be imposed by controlling forces or barriers and the police in real situation. This is done by introducing signals (decision thresholds or subjective judgements about pay-off and penalties) that can be transmitted from agent to agent in the crowd and from the crowd to the controlling forces. Using Bayes Nash Theory, the model analytically controls the thresholds to maintain unique equilibrium. The feedback technique is a practical alternative to Bayes where Dynamics model will be employed to enable simulation of cooperation between agents in the crowd and maintain an equilibrium which in actual situation corresponds to rights of individual for freedom of expression that must be controlled with minimal force. The basis for such idea is the Oka Crisis in 1990 and the APEC Protest in 1997 which behaviour over time graphs are compared and analyzed. The graph is a representation of crowd’s point of view and cumulative pay-offs awarded and penalties inflicted over time as shown below. Figure 7- Behaviour overtime graph for Oka crisis Figure 8 - Behaviour overtime graph for APEC The Atkeson’s Game-Theoretic Model provided the parameter definition such as signals, strength, positive payoff, negative penalty with resulting fraction for the crowd that can be controlled by the controlling forces. The behavioural overtime graph using Atkeson’s Game-Theoretic is shown below Figure 9 - Stable behavioural overtime graph The feedback implementation works like the following: When strength threshold s applied and given a small value, the rioters succeed in the event and vice versa. If the updated strength is greater that the strength thresholds of the crowd, the police will use force to control the crowd. Therefore, adjusting the police strength threshold downward will likely to result in penalty on the crowd since their strength is greater than the police. The exchange of feedback will then result to balance in the crowd’s cumulative gain as shown below. Figure 10 -Simulation mode with feedback implemented References: Donikian S, 2000, A programming environment for behavioural animation, Campus de Beaulieu, France, pp. 1-10 Kamel M, Fkry M, Mashat A, Biqami N, Barhamtoshy H, & Bedewy I, 2004, Monitoring, Surveillance and Control of the Crowds in the Holy Sites Using SCADA System, KAAU Computer Science Dept, Cairo University, pp. 1-11 Schubert J. & Suzic R, 2007, Decision support for crowd control: Using genetic algorithms with simulation to learn control strategies, Swedish Defence Research Agency, Third IEEE Workshop Situation Management, SIMA-3.4, pp. 1-7 Taylor I, Desai S, Stemate L, & Pasca C, 2007, Using Feedback to Conduct Crowd Control, Centre of Operational Research and Analysis/ Defence Research and Development Agency, Canada, pp. 1-20 Read More

nt in the simulation is considered anything, individual or group that has sensors and effectors representing the real-world agents such as rioters, peaceful protesters, and the police as shown below. Figure 3- Stochastic agent-based simulation 2. Behavioural Animation Model Important Details Model characteristic Description Comments Crowd modelling type Macroscopic/behavioural Crowd movement /interaction Crowd AI Crowd simulation technique Real-time agent based Real-time application capability Yes Real-time agent motion and crowd flow using 3D models and scenario.

Real-life characters with respective behavioural mechanism, bio-mechanical movements and gestures. Contribution to knowledge & technology Multi-model approach and introduction of GASP & HPTS Usefulness for future study/experiments -Generation of real-life characters - Database of agent states & perception - Algorithms for bio-mechanical movements and gesture - Real-life 3D characters can be used for other experiments using 3D modelling software - Data from databases can be use for future study such as states, perception, and algorithms The idea behind Donikan (2000) behavioural animation model is to simulate autonomous beings like people and organisms.

Similar to real-time agent-based simulation it is dependent on behavioural characteristics and forces in the environment – decision, action, communication, perception, and so on. However, this model is more complex since incorporates various models – environment, mechanical, motion control, sensor models, geometric models, and various scenarios. The main simulation environment is GASP or General Animation and Simulation Platform while HPTS or Hierarchical Parallel Transition Systems is used to model the behavioural patterns.

The state in which the agent is in is represented by parameters RANDOM, DELAY, WEIGHT, and TIMEGATE which is actually the transitional behaviour of the agent reacting to forces. The model generate realistic geometrical model through a 3D modelling system which walk through is possible in real time as shown by the model’s application and Virtual Museum below. The principle used in this model is to have real-life characters (attendant, thief, sculptures, photographer, mother and son, and so on) interacting with an agent (visitor) so life can observed as well as behaviour modification of characters as they are introduced to pre-emption mechanism , dynamics, and parameters mentioned earlier.

The simulation is main influence by Sense-Control-Action or SCA loops with accompanying planners. The loops determine the reflexive behaviour through a network of nodes interconnecting sensors and effectors. Through a database of states and perceptions, the agent selects its perception, plan, and action. The motion as shown in the figures above is humanoid motion control which consists of motion algorithms mimicking bio-mechanical movements and gestures. 3. SCADA System for Crowd Control Important Details Model characteristic Description Comments Crowd modelling type Macroscopic/closed loop Crowd movement /interaction Density count/space signature Crowd simulation technique Real-time monitoring/surveillance Real-time application capability Yes CCD Camera for real-time crowd monitoring, surveillance, and control.

Real-time density count using image background frame differencing and space signatures through agents form, dimension, and luminosity. Contribution to knowledge & technology Background frame differencing/ image fine and coarse pattern density modelling Usefulness for future study/experiments Pattern matching Frame differencing Space signature -Data can be use to match patterns in other scenario - Other study can use or improve (remove errors caused by objects with similar intensity) frame differencing technique and space signature which is somewhat innovative.

The SCADA system is generally intended for crowd monitoring, surveillance and control. It uses sensors, algorithms for automatic crowd estimation, and pedestrian control.

Read More