Collective Communication Strategies for Space Exploration

In this paper we compare the efficiency of three co-operation strategies that enable a simulated swarm to collaborate on a searching task. Communication between swarm entities takes the form of decentralized direct peer-to-peer messages. This is in contrast with most swarm research which seeks to mimic the indirect communication observed in nature. A software simulation was created using C#, swarm robots of various colours are placed on a 2D grid and given the task of finding and collecting items that match their colour. If they encounter an item that they cannot collect they send a message to their closest neighbours who then pass it on to their neighbours, robots that match the colour then respond to the help message and move to the item.
The co-operation strategies dictate how far the message will be sent and how many robots are allowed to respond, depending on the strategy the number of responders can be one or many robots. The first strategy permits an unlimited number of swarm robots to respond to a help message. However, the amount of time it takes to find all items can be negatively impacted due to the mass movement of the swarm to this one item. The second strategy limits the number of responders to a select few, the message is not passed on further if within the first message pass there are swarm entities that can respond to the message. The third strategy requires a back and forth dialogue in which only one responder is chosen by the robot sending the help message. The swarm uses Dijkstra’s algorithm to find the shortest path when navigating the grid, they are given random coordinates to move to, these are updated when the target is reached. When moving to the target they scan each grid cell for items, if an item is found it is either foraged or a help message is sent to other swarm members. The research seeks to compare the performance metrics of the different strategies. Most notably, it looks at how much overall effort is wasted responding to help messages and whether one strategy is ideal given a certain swarm size. Future research will seek to implement an Autonomic Manager which can reconfigure the swarm composition and communication strategy in real time.
Swarm communication techniques will become more relevant as future space missions progress from a monolithic rover to autonomous swarms. As the numbers of rovers increase it will be more difficult and time consuming to control each with navigational waypoints. The communication lag to and from Earth could prove costly if a rover encounters a hazardous situation or needs to request help from a rover that carries a specific instrument. Being able to act autonomously and collaborate on a task would speed up mission completion time and provide a more robust and adaptable model for exploration.