New Threat Monitoring System for Autonomous Vehicles?
January 25, 2013 in Technology
“Roke Manor Research Ltd (Roke), a Chemring Group company, has developed the world’s first threat monitoring system for autonomous vehicles that emulates a mammal’s conditioned fear-response mechanism. The STARTLE system uses a combination of artificial neural network and diagnostic expert systems to continually monitor and assess potential threats.”
“The STARTLE architecture is presented, which emulates the mammalian conditioned-fear response mechanism. A lightweight feed-forward threat monitoring task receives relatively unprocessed data from multiple sensors and can alert the system to change its attention from its assigned task, cueing a model-driven, goal-proving threat assessor that can divert sensing and processing resources to resolve whether a perceived threat to own platform is real. This information can then be used (by separate processing functions) to determine an appropriate response to that threat (i.e. fight, flight or ignore). A proof of concept demonstration has been implemented that employs an Artificial Neural Network to implement the threat detector and a diagnostic expert system to implement the threat assessor. The demonstration exercised STARTLE within a simulated vignette involving an uninhabited ground vehicle undertaking a route clearance mission. The software was implemented using the Roke NMSE modelling tool and demonstrates the core STARTLE components, including the generation of sensor re-cueing events during the threat confirmation process.”