Abstract

Joint organizational planning and plan execution in risk-prone environment, has seen renewed research interest given its potential for agility and cost reduction. The participants are often asked to quickly plan and execute tasks in partially known or hostile environments. This requires advanced decision support systems for situational response whereby state-of-the-art technologies can be used to handle issues such as plan risk assessment, appropriate information exchange, asset localization and adaptive planning with risk mitigation. Toward this end, this thesis contributes innovative approaches to address these issues, focusing on logistic support over risk-prone transport network as many organizational plans have key logistic components. Plan risk assessment involves property evaluation for vehicle risk exposure, cost bounds and contingency options assessment. Appropriate information exchange involves participant specific shared information awareness under unreliable communication. Asset localization mandates efficient sensor network management. Adaptive planning with risk mitigation entails limited risk exposure replanning, factoring potential vehicle and cargo loss. In this pursuit, this thesis first investigates risk assessment for asset movement and contingency valuation using probabilistic model-checking and decision trees, followed by elaborating a gossip based protocol for hierarchy-aware shared information awareness, also assessed via probabilistic model-checking. Then, the thesis proposes an evolutionary learning heuristic for efficiently managing sensor networks constrained in terms of sensor range, capacity and energy use. Finally, the thesis presents a learning based heuristic for cost effective adaptive logistic planning with risk mitigation. Instructive case studies are also provided for each contribution along with benchmark results evaluating the performance of the proposed heuristic techniques.