IEEE Systems, Man and Cybernetics Magazine - October 2023 - 28

{Vehicle States},
{Environment States},
{......}
Sensing Delay
(Output Delay)
Roadside Sensors
Cloud/Edge
Controller
{Expected
Acceleration},
{Expected
Steering},
{......}
Control Delay
(Input Delay)
Controlled
Connected Vehicle
Figure 1. The connected vehicle cloud control system.
Therefore, in this article, we concentrate on the problem
of cloud control of connected vehicles with distinct
communication-induced sensing delay and adopt a predictor-observer
structured control strategy for both known
and unknown delay cases. The main contributions of this
article could be summarized as follows.
1) Considering the linearized control dynamics of the connected
vehicle, the predictor-observer structured controller
is proposed for both known and unknown output
delay cases. For known delay cases, the stability condition
under a certain delay length and model uncertainty
is conducted in a novel manner based on the SBRL [25].
Meanwhile, the stability condition under a certain delay
range and model uncertainty is uniquely conducted for
unknown delay cases based on the LKF method. The
CCL-based [26], [27] iterative algorithm is also provided
to determine stabilized controller gains for both cases.
2) The proposed prediction-based delay compensation
strategies are verified through simulation built with
Simulink-CarSim, which contains a high-fidelity vehicle
dynamic model. The prediction performances of two different
strategies are evaluated and compared in detail,
including a control performance comparison with the
ordinary constant parameter controller.
This article is organized as follows. The " Cloud Control
of Connected Vehicle Under Sensing Delay " section
provides the modeling of the connected vehicle cloud
control system with sensing delay. The " Delay Compensation
Strategy and Corresponding Controller " section
presents the stability analysis for the linear output feedback
system under two different predictor-observer
structured controllers, corresponding to known and
unknown output delay cases. Algorithms to determine
the stabilized controller and observer gains are also presented.
Simulation results on the control of the connected
vehicle are provided in the " Simulation Experiments "
section. Finally, conclusions and future works are summarized
in the " Conclusion " section.
28 IEEE SYSTEMS, MAN, & CYBERNETICS MAGAZINE October 2023
Cloud Control of Connected
Vehicle Under Sensing Delay
As described in the preceding section, under
the ideal working mode, the centralized connected
vehicle cloud control system could collect
vehicle running states through distributed
network sensors, process the sensing information,
and generate control commands for the
vehicle to follow. Among all these processes,
delays might be involved with the network
sensing process and the control command distributing
process, as shown in Figure 1. These
two delay parts will negatively affect the system's
control performance. This study mainly
considers the sensing delay part since it will
be more prominent in real scenarios and
could be compensated distinctly with the prediction.
Besides, the proposed delay scenario
could also cover the cases in which the connected vehicles
adopt only networked sensing information and make control
decisions by onboard devices. Since connected vehicles usually
work under ordinary working conditions (compared
with limiting conditions), the coupling of the lateral and longitudinal
dynamics is weak in the scenarios concerned.
Therefore, for complexity issues, the control dynamics of the
connected vehicle could be decoupled into lateral and longitudinal
control, which could cover various control scenarios.
The dynamics of longitudinal and lateral control are introduced
separately in the following sections.
System Dynamics
The lateral control of the connected vehicle is basically to
control the steering to track a given trajectory. In a more
specific description, the positioning information of the
controlled vehicle is acquired by roadside sensors, and the
cloud-side controller utilizes the information to determine
the target steering angle to track a predefined trajectory.
The target steering angle is finally sent to the controlled
vehicle and executed through an onboard power steering
system. Based on the existing research on the lane keeping
system [28], the connected vehicle lateral control model is
introduced. By ignoring the impact of external disturbances,
such as wind and other factors, the lateral control system
can be modeled as shown, following discretization
using a zero-order holder:
xk 1 Ax kB kP kr
()
() () () ()++ (1)
()
+= dt
latlat
d =
=
xk ek ek ek ek112 2
lat
() (),( ), (),( )
kK xk
6 oo @T
(2)
where Alat is the state matrix of the discrete lateral control
system, Blat is the input matrix of the discrete lateral control
system, Pr is the feedforward input matrix of the lateral
control system, Klat is the feedback control gain, and x
represents the state vector of the lateral control system
and includes the following components: e1 is the distance
IEEE Systems, Man and Cybernetics Magazine - October 2023

Table of Contents for the Digital Edition of IEEE Systems, Man and Cybernetics Magazine - October 2023
