Calendar

Previous May Next Previous 2017 Next
Mo. Tu. We. Th. Fr. Sa. Su.
1 2 3 4 5 6 7
8 9 10 11 12 13 14
15 16 17 18 19 20 21
22 23 24 25 26 27 28
29 30 31 1 2 3 4

Affiliations

Links

Collaborative and distributed control in networked systems

Carlos Canudas de Wit
Federica Garin
Ruggero Fabbiano

Consensus of heterogeneous multi-agent systems

A new consensus algorithm for heterogeneous multi-agent systems is provided, proposing a control strategy based on a consensus algorithm which is decoupled from the original systems; consequently, its major advantage remains in the separation of the stability analysis of each subsystem and the distributed control algorithm. It is shown that our method allows using classical distributed consensus algorithms such as simple integrator consensus (with or without delay) and distributed consensus filter algorithms, and, for many multi-robot applications, it is interesting to impose a particular configuration for the robotic agents. We discusse the design and analysis of a distributed algorithm for the compact deployment of agents, where the behavior of each vehicle is only dependent on local information; the objective is to achieve the most compact formation possible and, to solve this problem, we propose, in a first step, two uncorrelated controllers: one designed for dispersion with connectivity maintenance, and a second designed to minimize inter-agent angles. An improved controller including variable gains, particularly designed to avoid singular configurations, is also provided. Lastly, we propose a sequential strategy composed of the two previously mentioned controllers and a stability analysis based on hybrid systems theory.

Source localization

Bi-dimensional projection of a 3D source-seeking simulation for an isotropic diffusion process

The problem of source localization consists in finding, by one or several agents eventually cooperating each other, the point or the spatial region from which a quantity of interest is being emitted. Source-seeking agents can be fixed sensors, that collect and exchange some information about the signal field and try to identify the position of the source (or the smallest region in which it is included), or moving devices equipped with one or more sensors, that phisically reach the source in an individual or cooperative way. This research area is attracting a rapidly increasing interest, in particular in applications where the agents have limited or no position information and GPS navigation is not available, as in underwater navigation or in cave exploration: for instance, source localization is relevant to many applications of vapor emitting sources such as explosive detection, drug detection, sensing leakage or hazardous chemicals, pollution sensing and environmental studies. Other fields of interest are sound source localization, heat source localization and vent sources in underwater field. Techniques present in literature either are based on a specific knowledge of the solution of the diffusion process, or make use of an extremum-seeking approach, exciting the system with a periodic signal so as to explore the field and collect enough information to reconstruct the gradient of the quantity of interest. Our approach lies in the computation of derivatives (potentially of any order) from Poisson integrals that, for isotropic diffusive source in steady-state, whose solution satisfies the Laplace equation, allows for a gradient search with a small computation load (derivatives are computed by integrals) and without requiring any knowledge of the closed-form solution, avoiding in the same time extremum-seeking oscillations; this has the additional advantage of an intrinsic high-frequency filtering, that makes the method low sensitive to measurement noise.

Demo

Distributed real-time simulation of numerical models

Vehicles and powertrains are complex systems where different domains are involved. The need of quick innovation in the automotive domain made simulation necessary at early stages of the development cycle, and representative phenomenological models of powertrains have been developed and have been used in the design phase under domain dedicated tools. However, their use for controls validation using model-in-the-loop (MIL) and hardware-in-the-loop (HIL) was prevented due to performance limitation of widely used single-solver/single-core simulation approaches. Multicore simulation for complex systems has been studied with a focus on simulation duration speedup, and the methodology of parallelization across the model has been selected for such problem where strong interactions between the model components are observed. The current study showed that decoupling the model parts by relaxing their data dependencies is promising in term of simulation speed (by increasing the parallelism) and results accuracy; besides, test results on engine model showed that, with the model partitioning, it is possible to use efficiently variable-step solvers thanks to the decrease of the number of discontinuities, so the number of integration interrupts, in each subsystem. Further work will investigate the combination of the use of variable-step solvers in split model with the use of multicore architecture for parallel computing, in order to improve the simulation speedup while keeping results accuracy under control.

Selected papers