A Domain Independent Pattern Recognition System to Support Space Mission Planning
Juan Manuel Delfa Victoria, Yang GAO, Nicola Policella, Alessandro Donati · Rare & Special e-Zone (The Hong Kong University of Science and Technology) · 2011
In the history of space exploration, the complexity of spacecrafts and missions have never stopped growing. This makes manual operations very challenging if not impossible. As a result, space agencies have started to move toward autonomous operation and mission planning systems for both on-ground and on-board scenarios. A planetary rover represents a typical complex mission, as the rover needs to directly interact with a dynamic environment that will introduce errors to any pre-designed plan while direct telecommanding from human operators is not possible due to communication delays, demanding more intelligence from on-board software. In future scenarios like a permanent base on the moon or a manned mission to Mars, robots and humans will need to collaborate closely using a common natural language. The operators should be in charge of defining high level orders to be executed by the robots, which will have to coordinate their actions in order to achieve all the goals with the best possible performance. The planner on-board each robot is the final responsible of the generation and maintenance of a valid plan, taking into consideration the set of goals assigned and the changing conditions of the robot and environment. The planner should as well be capable of generating its own high level goals in case of unexpected events.