A real-time Deep Neural Network (DNN) adaptive control architecture is developed for general uncertain nonlinear dynamical systems to track a desired time-varying trajectory. Lyapunov-based method leveraged develop adaptation laws the output-layer weights of DNN model in while data-driven supervised learning algorithm used update inner-layer DNN. Specifically, are estimated using an unsupervised provide responsiveness and guaranteed tracking performance with feedback. The trained collected...

A continuous adaptive controller is developed for nonlinear dynamical systems with linearly parameterizable uncertainty involving time-varying uncertain parameters. Through a unique stability analysis strategy, new feedforward term along specialized feedback terms, to yield an asymptotic tracking error convergence result by compensating the nature of Lyapunov-based shown Euler–Lagrange systems, which ensures and boundedness closed-loop signals. Additionally, function approximation converge...

We propose a deep reinforcement learning (DRL) methodology for the tracking, obstacle avoidance, and formation control of nonholonomic robots. By separating vision-based into perception module controller module, we can train DRL agent without sophisticated physics or 3D modeling. In addition, modular framework averts daunting retrains an image-to-action end-to-end neural network, provides flexibility in transferring to different First, convolutional network (CNN) accurately localize indoor...

This work explores the distributed state estimation problem for an uncertain, nonlinear, and continuous-time system. Given a sensor network, each agent is assigned deep neural network (DNN) that used to approximate system's dynamics. Each updates weights of their DNN through multiple timescale approach, i.e., outer layer are updated online with Lyapunov-based gradient descent update law, inner concurrently using supervised learning strategy. To promote efficient use resources, observer uses...

This letter presents a novel estimator and predictor framework for target tracking applications that estimates the pose of mobile intermittently leaves field-of-view (FOV) agent's camera. Specifically, uses an attention deep motion model network (DMMN) to estimate dynamics when is in FOV DMMN predict position, orientation, velocity outside FOV. A Lyapunov-based stability analysis performed determine maximum dwell-time condition on measurement availability, experimental results are provided...

Data-based, exponentially converging observers are developed for a network of stationary cooperative cameras estimating the Euclidean distance to features on moving object (and hence, objects' accurately scaled structure), without requiring typical positive depth constraint and only remain in one camera's field-of-view. The demonstrate that synthetic persistent view relative each camera is sufficient maintain estimates. A Lyapunov-based stability analysis demonstrates geometric approach...

A controller, estimator, and predictor framework is developed for tracking a moving target using network of mobile cameras, with non-overlapping fields-of-views operating regions. Using Lyapunov-based switched systems approach, the proposed proven to be robust intermittent feedback, estimates target's pose motion model are remain bounded, provided that minimum maximum dwell-time conditions satisfied (i.e., time must observed may unobserved, respectively). This allows teams cooperative agents...

We propose a deep reinforcement learning (DRL) methodology for the tracking, obstacle avoidance, and formation control of nonholonomic robots. By separating vision-based into perception module controller module, we can train DRL agent without sophisticated physics or 3D modeling. In addition, modular framework averts daunting retrains an image-to-action end-to-end neural network, provides flexibility in transferring to different First, convolutional network (CNN) accurately localize indoor...

A method is provided to enable a nonholonomic vehicle explore an unknown environment with intermittent state feedback. maximum dwell-time condition determined via Lyapunov-based switched systems approach maintain overall system stability despite the loss of feedback and presence external disturbances. minimum ensure tracking error converges within desired neighborhood trajectory. Utilizing proposed conditions, vehicle's remains globally uniformly ultimately bounded, enabling exploration...

A relay-explorer control method for nonlinear multi-agent systems is developed to allow a relay agent intermittently provide navigational feedback an explorer leader. distributed controller formation and leader tracking the followers, enabling system explore unknown environment indefinitely. To compensate lack or inability use sensors, state observers are used propagate estimates agents (e.g., in GPS-denied regions). Stabilizing dwell-time conditions determined via Lyapunov-based switched...

A continuous adaptive control design is developed for nonlinear dynamical systems with linearly parameterizable uncertainty involving time-varying uncertain parameters. The key feature of this a robust integral the sign error (RISE)-like term in adaptation law which compensates potentially destabilizing terms closed-loop system arising from nature Lyapunov-based stability analysis ensures asymptotic tracking, and boundedness signals.

Distributed state estimation is examined for a sensor network tasked with reconstructing system's through the use of distributed and event-triggered observer. Each agent in employs deep neural (DNN) to approximate uncertain nonlinear dynamics system, which trained using multiple timescale approach. Specifically, outer weights each DNN are updated online Lyapunov-based gradient descent update law, while inner biases offline supervised learning method collected input-output data. The observer...