its-elementary-dear-watson

For the robots to collect dust they need to know where to go. In a space station it is undoable to just let the robots roam free, as in without smart guidance, throughout the station. Therefore we need to embedded intelligence into the system. Intelligence for this is multiparted. Over-time all particles end in the air-vents, but because the robots work 24/7 this could be avoided and collected earlier. For this we need algorithmens. There are already a lot of known theories that describe the displacement and movement of (dust) particles, like Particle Swarm Optimization, Brownian Motion, Normal (Gaussian) Distribution. These theories probably do not fully apply to zero-g and that should be tested and adjusted in a lab environment,

Second part of the intelligence should apply to geospatial awareness and robotic sensory data. Most of these systems are already known in one way or another.

To prove viability of the concept we relied on having access to the raw sensor data. Unfortunatly we were unable access this data within the resources of the challenge. Our reasoning to the requirements for this data was to be able to get a deeper insight, such as behavioural data, recurring and hidden patterns of the robots and experimental space in conjuction with BlueMix Watson Analytics. For this reason we are not able to demonstrate our concept with relevant robotics sensor data. Still, we believe to have a very compelling concept on how to start managing habitats autonomously.

IMPACT & NEEDS We believe, in order to prove our concept, there is need for a lab to simulate the working of the algorithms and sensory robotic data. In both a physical and a virtual environment. In a bigger picture we see this system being used in different kinds of habitats, from earthbound, convinced and either in open space, evolving into a real Autonomous Habitat Manager.

Project Information

Resources