Rover routing for planetary surface exploration considering system layout trade-off and multiple stakeholders’ perspectives

Abstract

This thesis addresses the routing of planetary surface rover considering two important aspects associated with space exploration. The problem can be formulated as the vehicle routing problem with profits (VRPP). Given the base (landing location) and the set of sites that can be potentially explored, the VRPP determines the routes of a rover starting from a base, visiting selected sites to collect profits, and returning to the base, to maximize the sum of collected profits obtained at visited sites under constraints on a single route and the whole exploration mission. In the traditional VRPP framework, two important issues for the routing problem associated with space exploration - the coupling between the routes and rover’s layout trade-off and the various perspectives of multiple stakeholder groups - additionally considered. In the first part of the thesis, the routing optimization of planetary surface rover considering the core composition of the rover, which is one of its key design decisions. The rover system is simplified as two core subsystems having important functional trade-off - the mobility/operation subsystem and the science subsystem. A double-loop solution methodology that can determine the best weight distribution of two subsystems as well as the optimal routing for the given exploration map is proposed. Case studies for two sets of surface exploration instances having different characteristics are conducted to demonstrate the validity of the proposed optimization procedure. The second part of this thesis proposes a routing problem that involves the perspectives of multiple stakeholder groups. To reflect various benefit types obtainable from the planetary exploration, the concept of profit vector - rather than a scalar - assigned on a site. Each element of the vector represents the assessment of the site provided by a specific stakeholder. The objective for the routing optimization with vector profit is to maximize the sum of profits for the least satisfied stakeholder - sequentially in a lexicographical sense. A mixed-integer linear programming (MILP) formulation of the proposed problem referred to as the Vehicle Routing Problem with Vector Profit (VRPVP) is presented. A column generation based solution methodology that can find near-optimal solution for the VRPVP with optimality gap is developed. Two case studies - a planetary surface exploration and a Rome tour case - are presented to demonstrate the effectiveness of the proposed formulation and solution procedure.