Safety Diagnostics and Degraded Operational Modes for Off-road Unmanned Ground Combat Vehicles

Abstract

Unmanned ground combat vehicles are obligated to traverse safely to their destination in an extensive variety of hazardous situations along with challenging terrains and the objective of the paper is to aid this intent. In circumstances of restricted operations of the autonomous vehicle, the composite systems of the vehicle should ensure its safety. However, the current fleets of autonomous vehicles are lacking the ability to predict and operate as effectively as possible in a restricted operational domain. This paper proposes an approach to create safety diagnostics for unmanned ground combat vehicles by mainly depending on probabilistic predictions of critical situations. The predictions are achieved by a recursive Bayesian model along with constantly examining the changing environments which effect the perception sensor readings and the current behavior of the unmanned ground combat vehicles. To verify and validate the approach that this paper describes, the Mississippi State University autonomous vehicle simulator was used to run simulations of an autonomous vehicle affected by a fixed set of environmental parameters which were also used for computing the risk of failure using a Recursive Bayesian and Markov models. Finally, simulations are conducted for two scenarios to illustrate the effectiveness of the proposed approach.