MAgIC: Multi-Agent Intelligent Control of time-critical Cyber-Physical Systems over wireless

PROJECT OVERVIEW

TV news and newspapers abound with articles talking about smart energy grids, smart drones, smart factories, etc. These types of systems where a number of collaborating computational elements control time-critical physical entities, are referred as Smart Cyber-Physical Systems (CPSs). So far CPSs operators are still adopting XX century unscalable centralized control architectures and dedicated SCADA communication networks. We believe that a groundbreaking advancement in CPS can only happen with the development of autonomous distributed control over off-the-shelf Wi-Fi, a goal that nobody has yet achieved. This project aims at laying the foundation for pioneering architectures and algorithms, as well as a preliminary proof-of-concepts in the area of cooperative robotic manipulation over wireless, which is the CPS realm which requires the highest communication bandwidth according to the CPS challenge plane shown in the Figure. This project feature an interdisciplinary team with complementary expertise in Control, Communications, Robotics and Computer vision.

 

PEOPLE

Principal Investigator:

Luca Schenato

Multi-agent Control Systems

 

Members:

Angelo Cenedese UAVs Control	Stefano Ghidoni Robotics	Simone Milani Computer Vision	Federico Tramarin Industrial Communications	Roberto Oboe Mechatronics

 

Riccardo Antonello Control Engineer	Giulia Michieletto Post-doc	Enrica Rossi Ph.D. Student	Francesco Branz Post-doc	Matthias Pezzutto Ph.D. Student

External associates:

Ruggero Carli Control for robotics	Sfefano Vitturi Industrial automation

PROJECT OBJECTIVES

 

WP1) control over wireless: future CPS will necessarily run over wireless to exploit ad-hoc and rapid connectivity, however the inherent unreliability of the communication (packet loss, random delay) poses a formidable challenge. Despite the past 15-year extensive research on the topic, little penetration has appeared in industrial control applications. We believe that this is mainly due to two aspects: the first is that steady-state error distributions are heavy-tail, i.e. large errors are not rare. The second aspect is that current design procedures do not properly model realistic wireless protocols such as Wi-Fi.  

WP2) communication for control : current communications standards (as Wi-Fi, Zigbee) were not designed for control applications. In fact, throughput is not particularly relevant in control systems while both packet loss and delay negatively impact their performance. The current trend is to create new standards (WirelessHART, ISA 100.10a) which reduce to bare minimum both packet loss and randomness in the delay, at the price of an increased delay and the use of dedicated HW. In this project we will pursue a different approach by exploring the possibility to dynamically control the parameters available in the current Wi-Fi standards to dynamically change the rate of communication.

WP3) cooperative robotics over wireless: Cooperative robotics is a well established research field but it has been mainly focused on exploration and mapping such as SLAM where bandwidths of 10Hz are typically sufficient. In the context of robotic manipulation, cooperation over wireless is a more pristine field since it requires a bandwidth of around 1000Hz, which is two orders of magnitude faster of what Wi-Fi can reliably provide today. We intend to tackle this problem by designing novel distributed cooperative controllers robust to packet losses, which we believe could provide almost the same performance at a smaller bandwidth of 100Hz, and by boosting Wi-Fi bandwidth to 100Hz still guaranteeing small packet loss probabilities (around 10%). In order to validate the effectiveness of the theoretical tools developed in WP1-3, we intend to implement some of the proposed algorithms into four

WP4) proof-of-concept experimental demos: mobile vehicles and robotic manipulators integrated with Wi-Fi capabilities and on-board cameras.

 

 

 

SOFTWARE AND DATA

 

Wif-Fi transmission data logs used in [1] (.zip)

Wi-Fi4Control Experimental Packet Loss Sequences: (.zip)

 

PUBLICATIONS

 

Journal:

[J1] F. Branz, R. Antonello, M. Pezzutto, F. Tramarin, S. Vitturi, L. Schenato. Drive–by–Wi-Fi: Model–Based Control over Wireless at 1-kHz. (to be submitted at IEEE Transactions on Control Systems Technology), 20XX

[J2] F. Branz, R. Antonello, F. Tramarin, S. Vitturi, L. Schenato. Time-Critical Cooperative Networked Embedded Systems: feasibility and experimental assessment. IEEE Transactions on Industrial Informatics (under review)

[J3] M. Pezzutto, F. Tramarin, S. Dey, L. Schenato. Adaptive Transmission Rate for LQG Control over Wi-Fi: a Cross-Layer Approach. Automatica (under review)

[J4] E. Rossi, M. Tognon, R. Carli, L. Schenato, J. Cortes, A. Franchi. Cooperative Aerial Load Transportation via Sampled Communication. IEEE Control Systems Letters, vol. 4(2), pp. 277 - 282, 2020

[J5] N. Lissandrini, G. Michieletto, R. Antonello, M. Galvan, A. Franco, A. Cenedese. Cooperative Optimization of UAVs Formation Visual Tracking. Robotics, 8(3), 52, 2019

[J6] M. Pezzutto, S. Dey, L. Schenato. Heavy-tails in Kalman filtering with packet losses. European Journal of Control, (50), pp. 62-71, 2019 [url]

[J7] M. Pezzutto, E. Garone, L. Schenato. Reference Governor for Constrained Control over Lossy Channels. IEEE Control Systems Letters and CDC 19, vol. 4(2), pp. 271 - 276, 2020

Conferences:

[C1] M. Pezzutto, L. Schenato, S. Dey. Transmission Scheduling for Remote Estimation with Multi-packet Reception under Multi-Sensor Interference. IFAC World Congress 2020, 2020

[C2] F. Branz, R. Antonello, F. Tramarin, T. Fedullo, S. Vitturi, L. Schenato. Embedded systems for time–critical applications over Wi-Fi: design and experimental assessment. Proceedings of IEEE International Conference on Industrial Informatics (INDIN'19), 2019

[C3] G. Michieletto, A. Cenedese Formation Control for Fully Actuated Systems: a Quaternion-based Bearing Rigidity Approach. In 18th European Control Conference (ECC) (pp. 107-112), 2019

[C4] F. Branz, M. Pezzutto, R. Antonello, F. Tramarin, L. Schenato. Drive-by-Wi-Fi: taming 1kHz control applications over wireless. European Control Conference (ECC'19),

[C5] M. Pezzutto, F. Tramarin, L. Schenato, S. Dey. SNR-triggered Communication Rate for LQG Control over Wi-Fi. IEEE Conference on Decision and Control (CDC'18), 2018

 

PRESENTATIONS

 

Control over wireless: a unfinished journey (Luca Schenato), Control Days Workshop, Univ. of Padova, May 2019

Automatic Control over Wi-Fi: a cross-layer Approach (Luca Schenato),Workshop on Automotive: Cyber-Physical-Systems, Univ. of Modena, June 2018

Smart Multi-agent Control Systems over Wireless: Challenges and Perspectives, (Luca Schenato), Control Theory Seminars, DII, Univ. of Trento, January 2018

4 Lectures (A. Cenedese) at the "Foundations of Formation Control: Theory and Applications" IAAC workshop, Herzliya (Israel) - 11 Novembre 2019

On Formation Control for Multi-Agent Systems, (A. Cenedese) "Rigidity Theory for Multi-Agent Systems meets Parallel Robotics" Workshop, Nantes (France) - 30 Novembre 2018

FUNDED BY

 

Department of Information Engineering University of Padova