Development of a Swarm Robotics Model for Mining Robots
Developing a real world model for a swarm of mining robots by studying trade-off patterns between food intake and mortality risk in different natural swarms with focus on ants. In our latest model, we are using pheromone trails to guide robots towards the food source and analyze information propagation in case of dead ants.
[Update 09/24/2014: Paper on Foraging Trade-Off Behaviours accepted in Robio 2014. See publications.]
Multi-robot Fault Tolerant Area Coverage with Limited Communication and Sensing Ranges
Multi-robot coverage is an important problem in many applications, like floor cleaning, mine sweeping, mapping, surveillance, search, and rescue, etc.
This can by achieved by decomposing a region into cells or polygons and assigning each partition to an agent for coverage. But, sometimes these agents can fail during operation and the area of the faulty agent must be redistributed. If global communication was available then the agents can easily detect the fault and repartition the regions. However, agents have limited communication and sensing capabilities, hence, the faulty agent may not be detected and its area may not be covered.
We are working on the development of fault tolerant area coverage algorithms, that enables quick detection of the faulty agent under limited communication constraints and redistributes the area such that the complete area is covered. The approach is based on periodic communication between agents and scheduling of meeting time.
The analysis of the algorithm has been verified using simulations.