This program will develop an innovative software tool for optimizing control of heterogeneous swarming systems using Swarm Signal Temporal Logic (SSTL) specifications. The proposed solution will enable task prioritization and scheduling, and distributed optimization of heterogeneous multi-agent systems (MAS) with the ability to scale the agent team, environment, and degree of heterogeneity with minimal effects to the computational complexity. The SSTL framework enables swarm time- and spatial-based task definitions and scheduling. Methods for coordination and control of agent behavior, in terms of desired robustness and total energy consumption, will be implemented using game theoretics and gradient descent in a distributed manner. Inter-agent communication methods will be developed, based on the nearest-neighbor communications models for use in the logic framework. SSTL algorithms developed will be implemented into a set of MATLAB language modules capable of declaring mission and objective requirements based on definitions of agents, tasks, and scheduling. The SSTL has a broad application range, including sets of multiple spacecraft, unmanned aerial/ground vehicles, and maritime vessels coordinating and cooperating in a MAS. Agent properties, such as dynamics models, mass, charge level, energy efficiency, communication range, and control update rates will be modeled. Software demonstrations will be performed to illustrate the feasibility and capabilities of the temporal logic approach to control numerous types of MAS vehicles and applications using SSTL specifications. Simulation valuations will determine algorithmic performance in terms of control convergence, energy consumption, and rates of successful task completion. System use cases will be outlined via CONOPS development, detailing the potential utility and benefits of SSTL algorithms for NASA and commercial applications.
Applying SSTL to NASA programs involving MAS enables definition of complex mission tasks for heterogeneous swarms with strict requirements for objectives in terms of time, duration, and cardinality. Specific applications include cooperative scouting activities and path planning, autonomous multi-objective scientific exploration, and communications networking. Task prioritization, time- and event-based scheduling, and distributed convergence optimization via SSTL has broad applicability to swarms of space, aerial, sea, and ground vehicles.
The SSTL algorithms apply to a variety of MAS that benefit from coordinated task planning and distributed optimization. Non-NASA applications include commercial or defense UAS/UGV, seacraft, or space vehicles used for exploration, transit, or surveying. The SSTL framework is applicable to artificial intelligence, system design verification, linguistics, and task management systems.