Sina Afshari

Modeling and feedback control of color-tunable LED lighting systems

This paper presents a color-science-based approach to feedback control design of color-tunable LED lighting systems for smart spaces. The general design problem is posed as the minimization of a cost function consisting of metrics that capture light quality, energy consumption and human comfort. A linear light transport map is used for modeling and identifying the optical fingerprint of the room. The feedback control law is then derived based on the identified model through gradient-based optimization of the cost function. Finally, experimental results are presented to highlight the performance of the feedback control law in terms of (1) energy savings, (2) delivered light quality, (3) adaptivity to external disturbances (such as daylighting) and (4) human comfort.

Decentralized Feedback Control of Smart Lighting Systems

This paper presents a framework for designing controllers for self-commissioning smart lighting systems with plug-and-play capability. A class of decentralized feedback control methods is proposed for this purpose. Theoretical results for stability and convergence of the proposed algorithms are presented. Further, an automated self-commissioning algorithm is designed to minimize re-identification efforts necessary for the decentralized controller in case of a change in the lighting configuration (e.g. the addition of a new fixture to an existing space). The implementation of this algorithm demonstrates significant reduction in the commissioning effort. Finally, centralized, decentralized and consensus-based control algorithms are implemented on an experimental adaptive lighting testbed. The performance of the decentralized methods is shown to be comparable to that of the centralized controller.Copyright

Using computer games for hybrid systems controller synthesis

We propose a formal method for feedback controller synthesis using interactive computer programs with graphical interface (in short, computer games). The main theoretical tool used in this method is the concept of trajectory robustness, which is established using the theory of approximate bisimulation. Approximate bisimulation has been used to establish robustness (in ℓ∞ sense) of execution trajectories of dynamical systems and hybrid systems, resulting in trajectory-based safety verification procedures. We define control autobisimulation function (CAF), which is the analog of control Lyapunov function for approximate bisimulation. CAF is used to characterize the family of all feedback control laws, called admissible control laws, that result in a close loop system with an autobisimulation function. A computer game can then be used to construct safe and correct execution trajectory for a nominal initial state, and use the trajectory-robustness property to guarantee that the control law is also safe for other initial states in a neighborhood of the nominal initial state. As a result, a safe and correct feedback control law for a compact noncountable set of initial states can be obtained by playing finitely many games.

Short Paper:: The Smart Conference Room: An Integrated System Testbed for Efficient, Occupancy-Aware Lighting Control

A key challenge to lighting testbed development is the design of system architecture, both in terms of hardware and software, that supports the exploration of transformative systems concepts. Here, we introduce the Smart Conference Room (SCR) at the NSF Engineering Research Center for Smart Lighting. This regularly-used room combines multi-channel solid state light sources, advanced sensors, and sophisticated control algorithms to provide efficient and comfortable lighting where and when it is needed. We specifically discuss two interrelated sensing and control systems in the SCR. The first uses a sparse network of single-pixel color sensors for closed-loop feedback control of the light field in the room, driving the light field to a desired setpoint while harvesting incoming daylight and balancing energy costs. The second uses a low-resolution array of time-of-flight sensors to track occupants, identify their behavior, and trigger lighting modes, all while preserving their privacy. We describe ongoing and future experiments in the SCR that drive the specifications and development of new sensors, sources, and advanced control algorithms.

An optimization framework for control of non-square smart lighting systems with saturation constraints

Smart lighting systems are illumination systems that use feedback measurements from a network of color sensors to drive a set of spectrally tunable light sources to achieve a desired light field in an illuminated space. This paper proposes a general feedback control design framework for non-square smart lighting systems with saturation bounds, i.e., systems with a larger number of source channels (with limited maximum light output) than sensor channels. Since the number of inputs is larger than the number of measurements, a one-to-one setpoint based feedback control design is not possible because of the inherent redundancies. The feedback control design for such lighting systems is therefore posed as a constrained optimization problem with a cost function penalizing quality of light output and power consumed; and solved through a closed loop feedback approach. Two solutions to this problem are proposed: one based on Newton-Raphson method with projection and the second based on the interior point algorithm. We demonstrate the stability of the feedback loop for the projected Newton-Raphson method. The two proposed smart lighting algorithms are experimentally validated by implementation in a full-scale in-use smart conference room and a comparison of their performance is presented.

An adaptive smart lighting system

This demo abstract introduces an adaptive lighting testbed in the Smart Lighting Engineering Research Center located at Rensselaer Polytechnic Institute to highlight the advantages and capabilities of smart lighting systems. This testbed has been designed for experiments on a variety of topics such as color control, energy efficiency, self-commissioning, human factor studies, and localization. Section 2 describes the hardware and software used in the adaptive lighting testbed. Section 3 presents different types of experiments carried out using the testbed. The graphical user interface designed to monitor different performance criteria and adjust the design parameters is described in section 4.

A Plug-and-Play Realization of Decentralized Feedback Control for Smart Lighting Systems

This paper proposes decentralized feedback controllers with plug-and-play capability for a class of smart lighting systems. In these systems, each light fixture has a spectrally tunable light source and a multichannel color sensor, and no communication is required between fixtures. Further, the light transport in the illuminated space is diagonally dominant, i.e., the light sensed by each sensor is primarily from the source in the same fixture. The controller design problems are formulated and solved for two cases of decentralized setpoint tracking and decentralized quadratic optimal control (with tracking error and energy penalty). For the decentralized setpoint tracking problem, a mechanism is proposed to automatically determine the feedback gains using individual sensor measurements in a plug-and-play fashion without the knowledge of the light transport model of the illuminated space. For the decentralized quadratic optimal control, the design problem is approximated with a series of local optimization problems, which are solved in a decentralized manner using individual sensor measurements. A suboptimality bound for the solution of the approximated problem compared with the global optimum is obtained. The performance of the suggested controllers in terms of achievement of a desired setpoint and daylight harvesting for energy saving is validated and evaluated based on typical lighting parameters in a room-scale experimental testbed with light-emitting diode fixtures and color sensors.

LIGHT FIELD ESTIMATION AND CONTROL USING A GRAPHICAL RENDERING ENGINE

State-of-the-Art feedback control of lighting depends on point sensor measurements for light field generation. However, since the occupant’s perception depends on the entire light field in the room instead of the illumination at a limited set of points, the performance of these lighting control systems may be unsatisfactory. Therefore, it is critical to reconstruct the light field in the room from point sensor measurements and use it for feedback control of lights. This paper presents a framework for using graphical rendering tools along with point sensor measurements for the estimation of a light field and using these estimates for feedback control. Computer graphics software is used to efficiently and accurately model building spaces, while a game engine is used to render different lighting conditions for the space on the fly. These real-time renderings are then used together with sensor measurements to estimate and control the light field in the room using an optimization-based feedback control approach. We present a set of estimation algorithms for this purpose and analyze their convergence and performance limitations. Finally, we demonstrate closed loop lighting control systems that use these estimation algorithms and compare their relative performance, highlighting their benefits and disadvantages.