This article presents a new control approach and dynamic model for engineered flapping flight with many interacting degrees of freedom. paper explores the applications neurobiologically inspired systems in form Central Pattern Generators (CPG) to dynamics. A rigorous mathematical theoretic framework design complex three dimensional wing motions is presented based on phase synchronization nonlinear oscillators. In particular, we show flying dynamics without tail or traditional aerodynamic...

We study team composition in the context of a multi-player pursuit-evasion game between intruders and defenders. The has been previously studied assuming full state information on both teams. extend this problem by requiring defenders to detect using limited sensor footprint before pursuit can begin. simplify policy synthesis for heterogeneous decomposing perimeter defense task into patrol subtasks each performed homogeneous team. derive nonlinear relationship robot capabilities, sizes,...

This work studies a dynamic, adversarial resource allocation problem in environments modeled as graphs. A blue team of defender robots are deployed the environment to protect nodes from red attacker robots. We formulate engagement discrete-time dynamic game, where can move at most one hop each time step. The game terminates with team's win if any location has more than time. goal is identify strategies, well conditions that determines winner: graph structure, available resources, and initial...

The problem of constrained Markov game has recently attracted interests in the study multi-agent reinforcement learning (MARL). existing literature focused on safe MARL problems where safety constraints are imposed for each agent individually. In this work, we consider potential (MPG) with a shared constraint, cost function respect to constraint depends states and joint actions all agents. We adopt primal-dual framework tackle establish Slater condition ensure strong duality. Moreover,...

We consider a variant of the target defense problem where single defender is tasked to capture sequence incoming intruders. The intruders' objective breach boundary without being captured by defender. As soon as current intruder breaches or gets defender, next appears at random location on fixed circle surrounding target. Therefore, defender's final end game becomes its initial for game. Thus, players pick strategies that are advantageous well future games. Depending information available...

We formulate a scenario in which an autonomous defender is tasked with intercepting intruder that tries to reach circular target region. This variant of the defense problem proposed by Isaacs as pursuit-evasion game. Unlike original guarding and its various extensions, we consider effect partial information imposing sensing limitation robots. analyze game decomposing it into three phases: deployment, asymmetric information, engagement phase. Focusing on particular parameter regime, propose...

A scenario is considered in which two cooperative Attackers aim to infiltrate a circular target guarded by Turret. The engagement plays out the two-dimensional plane; holonomic have same speed and move with simple motion Turret stationary, located at circle's center, has bounded turn rate. When Turret's look angle aligned an Attacker, that Attacker neutralized. In this article, we focus on region of state space, wherein only one able reach circle—and even then, help its partner Attacker....

We consider a path guarding problem in dynamic Defender-Attacker Blotto games (dDAB), where team of robots must defend graph against adversarial agents. Multi-robot systems are particularly well suited to this application, as recent work has shown the effectiveness these related areas such perimeter defense and surveillance. When designing defender policy that guarantees path, information about adversary environment can be helpful may reduce number resources required by achieve sufficient...

This article presents a new control approach for engineered flapping flight with many interacting degrees of freedom.This paper explores the applications neurobiologically inspired systems in form Central Pattern Generators (CPG) to generate wing trajectories potential MAVs.We present rigorous mathematical and theoretic framework design complex three dimensional motions wings.Most demonstrators are mechanically limited generating trajectories.Because CPGs lend themselves more biological...

As robotic swarm systems emerge, it is increasingly important to provide strong guarantees on energy consumption and safety maximize system performance. One approach achieve these through constraint-driven control, where agents seek minimize subject a set of task constraints. In this paper, we an equivalent sufficient necessary optimality condition for energy-minimizing agent with integrator dynamics that only depends the state control actions agent. particular, show must have continuous...

A scenario is considered in which two cooperative Attackers aim to infiltrate a circular target guarded by Turret. The engagement plays out the dimensional plane; holonomic have same speed and move with simple motion Turret stationary, located at circle's center, has bounded turn rate. When Turret's look angle aligned an Attacker, that Attacker terminated. In this paper, we focus on region of state space wherein only one able reach circle - even then, help its partner Attacker. Runner...

In this paper, a zero-sum differential game is formulated and solved in which mobile Evader seeks to escape from within circle at whose origin lies stationary, turn-constrained Turret. The scenario variant of the famous Lady Lake shore-constrained Pursuer has been replaced with As former, it assumed that Turret's maximum angular rate greater than linear velocity Evader. Since two outcomes are possible, Game Kind arises - either wins by reaching perimeter circle, or Turret aligning latter's...

This paper is a companion to Chung 1 and explores the applications of neurobiologically inspired control systems in form Central Pattern Generators (CPG) flapping flight dynamics.We introduce two-layer CPGs mimic current hypotheses mammalian studies.It shown that symmetry breaking initiate recover from turning maneuver an effective strategy.Attempts at dissociating slow dynamics are preliminary comparisons wing motions between biological fliers artificial CPG networks made.

This paper investigates the control of phase difference in between three different motions bat flight: pitching, flapping, and lead-lag.For active control, a robotic test bed capable simulating wing is used to these differences using central pattern generators (CPG's).Previous work with expanded upon by modifying allow for dimensional entire bat, instead only wings.This done mounting onto 3D pendulum.Experiments analyzing steady state behavior bat's flight varying showed change pitch while...

Reinforcement learning (RL) approaches can illuminate emergent behaviors that facilitate coordination across teams of agents as part a multi-agent system (MAS), which provide windows opportunity in various military tasks. Technologically advancing adversaries pose substantial risks to friendly nation’s interests and resources. Superior resources alone are not enough defeat modern complex environments because create standoff multiple domains against predictable doctrine-based maneuvers....

This paper investigates obstacle-free simple motion pursuit-evasion problems where the pursuer is faster and game termination point capture. It well known that interior of Apollonius Circle (AC) evader's dominance region, however, it was unclear whether evader could reach outside initial AC without being captured. We construct a pursuit strategy guarantees capture an within arbitrarily close neighborhood AC. The derived by reformulating into nonlinear control problem, guarantee holds against...

This paper investigates a partial-information pursuit evasion game in which the Pursuer has limited-range sensor to detect Evader. Given fixed final time, we derive optimal strategy for Evader maximize its distance from pursuer at end. Our analysis reveals that certain parametric regimes, Evasion involves 'risky' maneuver, where Evader's trajectory comes extremely close pursuer's sensing boundary before moving behind Pursuer. Additionally, explore special case can choose time. In this...

This paper studies the synthesis of an active perception policy that maximizes information leakage initial state in a stochastic system modeled as hidden Markov model (HMM). Specifically, emission function HMM is controllable with set or sensor query actions. Given goal to infer from partial observations HMM, we use Shannon conditional entropy planning objective and develop novel gradient method convergence guarantees. By leveraging variant observable operators HMMs, prove several important...