eduzhai > Applied Sciences > Engineering >

Asynchronous Multi Agent Active Search

  • Save

... pages left unread,continue reading

Document pages: 27 pages

Abstract: Active search refers to the problem of efficiently locating targets in anunknown environment by actively making data-collection decisions, and has manyapplications including detecting gas leaks, radiation sources or humansurvivors of disasters using aerial and or ground robots (agents). Existingactive search methods are in general only amenable to a single agent, or ifthey extend to multi agent they require a central control system to coordinatethe actions of all agents. However, such control systems are often impracticalin robotics applications. In this paper, we propose two distinct active searchalgorithms called SPATS (Sparse Parallel Asynchronous Thompson Sampling) andLATSI (LAplace Thompson Sampling with Information gain) that allow for multipleagents to independently make data-collection decisions without a centralcoordinator. Throughout we consider that targets are sparsely located aroundthe environment in keeping with compressive sensing assumptions and itsapplicability in real world scenarios. Additionally, while most common searchalgorithms assume that agents can sense the entire environment (e.g.compressive sensing) or sense point-wise (e.g. Bayesian Optimization) at alltimes, we make a realistic assumption that each agent can only sense acontiguous region of space at a time. We provide simulation results as well astheoretical analysis to demonstrate the efficacy of our proposed algorithms.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×