Artificial Being (ABe)

ABe is visualized as a collection of mobile robots (or nodes), sensors, appliances and gadgets that together are networked and form the being. Though they are physically separable, a set of mobile agents that roam within these entities form the cells that nourish each of them with valuable nuggets of information gained by experience from the other, thus unifying them as the being. The robots depend on the mobile agents to serve them the code (programs) that will help them execute tasks, refine the code based on the results of execution and then share the same with other robots and entites that comprise the beingABe is thus a Cyber-Physical System.

ABe’s realization will thus provide a platform/test-bed for real-world distributed networked and autonomous robotics. It will mean that one will be in a position to actually program mobile robots to perform in a distributed manner.

The first challenge of developing the middleware for networked robotics calls for  the  design  and  development  of  a  software  framework  that  can  facilitate interactions  amongst  the  distributed,  interoperable,  heterogeneous  robotic entities  that  comprise  the  network  and  the  simplification  of  complex  robot control  software  systems  which  in  turn  can  ease  the  associated  application development  process.  In ABe mobile  agents  form  virtual  machine  based paradigms  and their associated  platform provides as the middleware.  A framework  that  exploits  key  features  of  mobile  agents (mobility, cloning, autonomy, etc.)  has been used within this middleware.

The  architecture of ABe is  mainly inspired by the Biological Immune Systems (BIS)  whose metaphors justify the use of mobile agents. Agents are looked upon as immune cells while the  robotic  nodes  form  metaphors  for  the  organs of  the  complex network that forms the ABe. The mobile agents (immune cells) migrate within the ABe (from  one  robotic node  to  another)  searching for nodes in need of the  services that they carry  (antigen  detection) as payload and provide the same to them (neutralization), thus empowering the associated robot to perform the given task (triggering  the  complement  system). The robots thus need not initially possess the programs for the tasks they are asked/required to execute. This in turn allows novice robotic programmers to tether their robots and make them pheromone for (robotic) programs that they require.

ABe also uses  bio-inspired paradigms such as pheromone diffusion and stigmergic sensing to control the clone population within the network. Unlike conventional ant pheromone laying mechanisms, the robotic nodes spread (diffuse) a virtual fragrance (pheromone) to attract the concerned mobile agents towards themselves. Mobile agent cloning aids in faster and on-demand transportation of payloads across the network while also effectively utilizing available bandwidth.

While we initially used AgentSpace for studying different algorithms for mobile agent migration for ABe, we are now using Typhon which has been developed in-house for emulating the same over a real network of nodes.


Comments are closed.