Will Schedulers be Available on Board in the Next Generation of Robots

摘要

This paper presents a simulated annealing solution procedure for a Resource Constraint Project Scheduling Problem (RCPSP) as it may be found in future space robotic applications. Such RCPSPs also are found in some game problems in which it is easy to compare computerized solutions with human play. The player, or the computer, has to find the best sequence of activities in which an objective function is maximized within a given time. The scheduling problem found both in robotics applications and in such games, is more difficult than usual because it is subject to temporal constraints, and to constraints for renewable, nonrenewable, and cumulative resources. The cumulative resources are depleted or replenished over time, depending on the player's choice. In this study, a simulated annealing algorithm is successfully tested. The results show that its performance can satisfy a good player and that it may be of some help in robotic applications. The results also show that the algorithm is robust because the heuristic is always able to find a rather good solution, and that it is realistic since the computing power needed for it will be available on board in the medium-term future.