Multi-agent systems such as multi-robot systems or vehicle platooning rely on wireless communication and perceptual sensing modalities that are known to be vulnerable to stealthy adversarial perturbation and intrusions. Such adversaries are deleterious to the system's stability and challenging to detect preemptively. In this research, we leverage information redundancy and conformity to the network-level dynamics of the system to design distributed and decentralized frameworks capable of detecting and discarding transmitted adversarial data and non-cooperative agents (robots), allowing for resilient decision-making, information consensus, and cooperation.
Bahrami, M. and Jafarnejadsani, H.
Distributed Detection of Adversarial Attacks for Resilient Cooperation of
Multi-Robot Systems with Intermittent Communication.
[ Under Revision]
[Request]
[ GitHub]
Bahrami, M. and Jafarnejadsani, H. (IEEE ICUAS 2022)
Detection of Stealthy Adversaries for Networked Unmanned Aerial
Vehicles.
The 2022 International Conference on Unmanned Aircraft Systems (ICUAS),
Dubrovnik,
Croatia, June, 2022.
[IEEE
Xplore],
[ arXiv]
[ GitHub]
Bahrami, M. and Jafarnejadsani, H. (IEEE CDC 2021)
Privacy-Preserving Stealthy Attack Detection in Multi-Agent Control
Systems.
The 60th IEEE conference on Decision and Control, Austin, Texas, USA,
December, 2021.
[IEEE
Xplore],
[ arXiv]
The human body has evolved to work with specific blood flow and blood pressure through
various feedback mechanisms that allow for adaptations to maintain blood flow and pressure
at adequate levels. For example, in cases of acute bleeding, blood pressure decreases due
to blood volume loss. Then, Reflex tachycardia, as a physiological response in the form of
a feedback mechanism, signals the brain to increase heart rate, compensating for a drop in
blood pressure.
Taking inspiration from such biological feedback mechanisms, we proposed a reconfigurable
fault-tolerant control and monitoring framework for electro-hydraulic systems to address
performance-degradation problems, particularly leakage faults that cause a pressure drop.
The framework incorporates two simultaneous feedback mechanisms designed to maintain
system performance and stability under faulty conditions. More specifically, we develop an
adaptive monitoring algorithm (a higher-order sliding mode observer) that allows for the
estimation and reconstruction of the leakage faults that cause a pressure drop and
consequently performance degradation in the systems. The second set of adaptive
feedback mechanisms leverages the signals generated by the monitoring algorithms to adjust
and regulate the system's supply pressure and the control algorithm.
These adaptive feedback-control algorithms enable the system to continue operating
normally despite the presence of leakage faults.
M. Bahrami, M. Naraghi, M. Zareinejad.
"Adaptive super-twisting observer for fault reconstruction in
electro-hydraulic systems." ISA transactions 76 (2018): 235-245.
[ScienceDirect],[
arXiv]
M. Bahrami, M. Zareinejad,
"Adaptive higher-order sliding mode observer for a class of nonlinear
systems", [In Preparation]
[Request]
M. Bahrami, M. Zareinejad.
"Bio-inspired fault-tolerant control in hydraulic actuators", [In
Preparation]
[Request]
Nature can be the most impressive source of inspiration for engineers in making more capable systems. Across nature, the movement of some animals strongly depends on their environment - a form of mobility known as passive locomotion. In this type of mobility, the environment ( air or water) serves as a propulsion as its flow causes the animal movement in the same direction. For instance, jellyfish and flatworms just swim along the flow. The blue crab and continental-shelf fish such as plaice can sense the direction of a tidal flow whether it is favorably directed or not, to use it for increasing their speed of swimming. In engineering systems, the hydraulic actuators are usually used in interaction with their environment to exert torques and forces. Sometimes, the direction of the external load exerted by the environment is the same as the actuator’s desired movement. The main objective of this research is to exploit the external load of the environment as a propulsion in position control. In such conditions, the hydraulic system finds a secondary supply of energy that leads to increased efficiency. To this aim, a new configuration of control valves is used to have a controllable supply pressure regulating the system pressure according to the actuator's demands. In the encounter with assistive loads, the system pressure decreases and the external load is used to move in the desired direction. A switched H-infinity control strategy is used to synthesize suitable control algorithms for position control in the presence of model uncertainties.