A Multiple Level of Detail Approach to the Tactical Movement Problem

The tacticalmovement problemis considered to be one in which a robotic agent is required to move around its world to complete a task. This agent has manipulation abilities which allow it to perform work on its local surroundings. The coupled optimisation of the agent movements and manipulations is thus of key importance to minimise the cost of completing the task. The driving practical application in this paper is one of cost effective excavation in a mining environment. The agent is a mining shovel and it has the ability to manipulate the world through excavation actions. The problem becomes one of determining the optimal path that the shovel should take and the dig operations that should be completed at each point along the path. An initial solution is presented to automatically generate an optimized dig plan for a large robotic excavator. A wavelet based detail reduction approach is used which allows a near optimal solution of the problem to be generated in practically useful timeframes.