Inter-agent communication graph that sense distance to the environment boundaries and to other agents within unobstructed lines of sight; (A2) The agents do not know the entire environment and their positions in it; (A3) Depending on the problem at hand, the guards are also allowed to exchange information with agents within line-of-sight through an asynchronous communication channel with delays and packet losses. This communication graph is depicted on the side; (A4) The agents are assumed to

more »

... olve asynchronously, i.e., a different sensing/communication/control schedule is allowed for each agent; (A5) For simplicity's sake, we model these agents as point masses with first-order dynamics. Assumptions (A1) through (A5) characterize what we refer to as visually-guided agents. Illuminating art galleries via incremental partition and deployment Combining the discussion in the earlier subsections, we obtain the following version of the Art Gallery Problem: starting from arbitrary positions, how should the agents move (and what should they communicate) in order to reach final positions such that each point of the environment is visible to at least one agent. This is what we refer to as the distributed art-gallery deployment problem. Remarkably, the difficulty of this problem is inherently due to the communication and sensing constraints: the agents are not given a map of the environment and no central entity controls them. The proposed algorithms allow for sensor-based, distributed, asynchronous execution and guaranteed visibility is achieved when the number of agents is at least ⌊n/2⌋. The algorithm is organized in three steps: