Page 31 - Read Online

Page 258 Sun et al. Intell Robot 2023;3(3):257-73 I http://dx.doi.org/10.20517/ir.2023.17

1.INTRODUCTION
In the past few decades, the autonomous ground vehicles (AGVs) are rapidly developing because of their wide
applications in military and civilian fields [1–4] . Vehicle motion control, as the main function for fulfilling
tasks, has become increasingly important in the design of AGVs. Since the sensors, controllers, and actuators
are interconnected through communication networks, AGVs have evolved into typical networked control sys-
tems (NCSs) [5,6] . Due to the limited bandwidth in-vehicle network, the network-induced delays and packet
dropouts impose additional constraints on the control loop in AGVs. Therefore, the control issues that con-
sider information exchange schemes have gained popularity in designing of AGVs [6–9] .

Path following control, which involves ensuring that a vehicle tracks the desired path with zero-steady state
error, is a fundamental ability for AGVs [10,11] . It is worth noting that the GPS, which is used for positioning
and navigation, is important to the path following control of AGVs. Up to now, it should be mentioned that
the advent of shared communication brings some new challenges due to the networked integration of path
following control. Therefore, the path following control needs to consider not only the in-vehicle communi-
cation network but also the out-communication network. In a traditional way, although many efforts have
been dedicated to developing path following control strategies, only a few of them can cope with time delays
or packet dropouts caused by networks due to the low networked integrations for AGVs [12,13] . Various con-
trol strategies have been developed to deal with such network-introduced difficulties, such as fuzzy control [14] ,
[8]
MPC algorithm , robust control method [9,15,16] , and adaptive control strategy [17] . However, the results of
those methods show some conservations in communication scheduling. Based on the observation of exist-
ing works, there is still much room for further research on the co-design of control and communication for
networked AGVs. This motivated us to study event-triggered path following control of AGVs.

To address the co-design issues of path following control and communication, the event-triggered communi-
cation (ETC) scheme, which transmits its measurement data only according to control requirements, has been
proposed and received much attention from the control community. Compared to traditional time-triggered
communication schemes, the ETC scheme is efficient in reducing communication frequency, thereby helping
to avoid network congestion in in-vehicle networks. In addition, a larger communication interval can provide
an extra time slot for scheduling vehicles among AGVs while preserving control performance [18] . However,
theevent-triggeredcontrolschemeofpathfollowingcontrolinAGVsisconsideredjustinafewofworks [19–22] ,
making it meaningful to develop ETC schemes specifically in path following control for AGVs. Furthermore,
since functional safety is very important to the motion control of AGVs, a more intelligent ETC scheme should
be well developed for path following control. In fact, ETC schemes have been extensively investigated in NCSs
due to their resource-saving nature [23,24] and the references therein. Up to now, several novel ETC schemes
have been developed to solve the problems arising from networks. For example, the adaptive ETC scheme
that depended on state gradient is proposed to adjust the event threshold in [25] , while a dynamic ETC scheme
that introduces an additional positive term in the traditional ETC scheme is presented in [26] . Resilient ETC
schemes are also proposed in [27,28] to deal with denial of service attacks, and learning-based ETC schemes
are developed to cope with deception attacks in [19] . Furthermore, a memory-based ETC scheme is devel-
oped to improve control performance in [29] . Although the existing works on event-trigged control strategies
show a flexible capacity to adjust transmission frequency, such previous works on the design of advanced ETC
need extra terms or dynamics to adjust the event-triggered parameters rather than the parameters themselves.
Therefore, these adjustment strategies of event-triggered threshold are blind to state perception, which splits
the co-design of communication and control to some extent. In addition, due to the limited bandwidth for the
in-vehicle communication network and limited capacity of calculation for the electronic control unit (ECU),
using ETC to reduce communication and control burden is highly advantageous for path following control of
AGVs.

Due to the limitation of the previous works, the state-measurement-based event-triggered control scheme is