ABe is visualized as a collection of mobile robots (or nodes), sensors, appliances and gadgets that together form a networked Artificial 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, thereby unifying them as the Being. The robots/hardware depend on a population of mobile agents that serve them with the code (programs) that will help them execute tasks, refine the code based on the results of their execution and then share the same once again with other robots and entities that comprise the Being. ABe is thus, an Intelligent 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 we will be in a position to actually program mobile robots to perform in a distributed and decentralized 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 serves as the middleware. A framework that exploits key features of mobile agents (mobility, cloning, autonomy, on-the-fly programming, 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. Each of these nodes/entities could need some kind of service/program/information – meaning that the node is attacked by an antigen. If you wish to see an animated version of this explanation, Click Me and then on the video below the title in that page.
The mobile agents (immune cells/antibodies carrying services/programs/info.) migrate within these networked entities comprising the ABe (from one robotic node to another) searching for nodes in need (trying to detect such antigenic attacks) of the services (antibodies) that they carry 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.
To realize ABe we use a multi-mobile agent platform, nicknamed Tartarus, developed in-house for emulating the same over a real network of nodes. Check out more on Tartarus here.
Hang-on, we will be releasing a new Python version of Tartarus, named Tarpy. Do visit these pages again for the same.