Scenario : setting crowd events using augmented reality and artificial intelligence

Abstract

SCENARIO aims to provide tools to aid organisers and attendees of crowd events in the organisation of the layout of attractions, reducing crowd congestion and enhancing the event attendees' safe experience. SCENARIO Desktop allows organisers to plan event layouts using its optimiser, and visualise the effect of such a layout on crowd flows using its simulator. The optimiser uses evolutionary AI techniques, in the form of multi objective genetic algorithms to solve a variant of an NP-Hard problem, the Facility Layout Problem (FLP). In contrast to classical FLP, SCENARIO also caters for areas containing obstructions. The simulator uses an agent based approach with the implementation of efficient pathfinding and spatial data structures and algorithms. Pathfinding uses global routing through hierarchical methods, and local routing through collision detection and nearest neighbour querying techniques. A variety of spatial structures reduce the simulator's space and time complexity. These include efficient structures to describe terrain and agent location, and provide rapid lookup capabilities. SCENARIO Bridge links the organiser and mobile applications. It converts layout and event information exported by SCENARIO Desktop in one format into a mobile device's native format, allowing crucial decoupling and scalability. SCENARIO Mobile is an Android app that visually lays out an optimised path for attendees to their chosen attraction. It uses pathfinding structures created by SCENARIO Desktop, GPS based locations and Augmented Reality (AR). Use of accelerometer and magnetic sensors ensures stability. The layout optimiser's convergence time varied from a few minutes to over an hour, with satisfactory layout output. The simulator ran at efficient rates. The majority of agents kept their free moving or attraction viewing states. The bridging system successfully produced the necessary files for use by the mobile application. Within the hardware limitations of GPS accuracy the routes displayed using AR were correct basing themselves on pathfinding structures used by the desktop application.