Dynamic Resonant-Inductive Wireless Power Transfer System for Automated Guided Vehicles with Reduced Number of Position Sensors

Abstract

This paper deals with the position detection of automated guided vehicles (AGVs) in dynamic resonant-inductive wireless power transfer (WPT) systems. A position detection is necessary to activate the correct transmitting coil. One of the simplest and most effective approaches for a position detection method is to use optical or magnetic position sensors for each coil. However, due to needing a high number of sensors, this technique is relatively expensive. Therefore, an AGV position detection technique based on a reduced number of optical or magnetic sensors (by a factor of two) is proposed. The proposed detection technique was verified experimentally by using a scaled-down prototype of the dynamic WPT system. The proposed approach can be easily implemented by uploading a specific program code to a microcontroller. The microcontroller with the code developed by us was used for processing data from AGV position detection sensors, activating a suitable transmitting coil and controlling an inverter of the dynamic WPT system. As shown by the experiments, due to the proposed approach for the position detection of AGVs and activation of transmitting coils, the number of the position detection sensors is reduced by a factor of two, leading to reductions in the overall cost and level of complexity of the dynamic WPT system without degrading its performance.

1. Introduction

With an ever-increasing demand for automation, automated guided vehicles (AGVs) have become integral parts of almost all modern warehouses or manufacturing facilities [1,2,3,4]. They are often used to transport goods from one place to another place within a warehouse. In contrast to traditional electrical vehicles used outdoors, AGVs used in warehouses have much lower speeds (<2 m/s) [1]. Since an AGV is a type of electric vehicle, it uses electric power for the normal operation of its motor, sensors, and embedded electronic devices. Therefore, a battery is usually installed in the AGV to provide power to it. However, batteries have some disadvantages: they increase the weight and volume of the AGV, they should be recharged over time, and their number of charge/discharge cycles is rather limited. Additionally, Li-ion batteries may cause fire if used improperly [5]. A very convenient solution to reduce the sizes of batteries or even eliminate them is to dynamically transfer electric power to AGVs (or other types of electrical vehicles) without using wires [6,7,8,9,10]. The technique is known as dynamic wireless power transfer (WPT). If this technique is used, an AGV can operate continuously without the need for stopping its operation. Therefore, the dynamic WPT increases productivity significantly within a warehouse or a manufacturing facility [7].

Dynamic WPT can be implemented by using either resonant-inductive or capacitive WPT techniques. The resonant-inductive dynamic WPT can be implemented by using either a long loop of wire at the primary (transmitting) side or a long array of multiple discrete coils at the primary side [8,11]. When AC currents pass through the long loop of wire or the multiple coils, a magnetic field is generated and then is picked up by the secondary (receiving) coil attached to the bottom side of an electrical vehicle. Due to the fact that the long loop of wire should be energized continuously, it emits large radiated electromagnetic interference (EMI) and dynamic WPT systems suffer from low efficiency (because of a low magnetic coupling coefficient between the primary and the secondary sides of the dynamic inductive WPT system and because of relatively large standby losses) [8,10,11]. In a multiple-coil configuration, only one primary coil (or several adjacent primary coils) having sufficient coupling with the secondary coil is energized (activated) [8,12]. Advantages of the dynamic resonant-inductive WPT systems based on this approach are considerably higher system efficiency and significantly lower EMI than those of the dynamic resonant-inductive WPT systems based on long-loop primary couplers [10,11].

The transmitting side of a conventional resonant-inductive dynamic WPT system (see Figure 1) based on an array of N coils usually consists of an inverter, a compensation capacitor (or several compensation capacitors connected in parallel to increase current ratings), and N electronically controllable switches (one for each coil) with a suitable control circuit [11,13] that can be based on a microcontroller. The switches are used to activate only one transmitting coil above or in the vicinity of which the receiving coil is located. In some dynamic resonant-inductive WPT systems, several adjacent coils may be activated simultaneously when an electrical vehicle is above or in the vicinity of them. If a dynamic WPT system is connected to AC grid, a power factor corrector (with DC output) needs to be connected between the AC grid and the dynamic WPT system DC input to maintain the almost-unity power factor and low total harmonic distortion of the input current.

In order for an electrical vehicle position to be detected accurately (for the activation of the correct transmitting coil), often, position sensors are used. Usually, they are placed near each transmitting coil (Figure 1). The sensors can be magnetic (e.g., tunneling magnetoresistive) sensors [14], ultrasonic sensors [15], or sensors based on one or more detection coils created within each transmitting coil [6]. Although they have not been shown for resonant-inductive dynamic WPT systems, optical sensors for electrical vehicle position detection have been used in capacitive dynamic WPT systems [7]. The detection of the position of an electrical vehicle by using magnetic or optical sensors is the simplest solution.

The main disadvantage of using sensor-based position detection is that a large number of position sensors is required (especially in case of long AGV paths). Using a large number of sensors increases the cost and complexity of the position detection circuit. Let us consider an example: the length of the AGV path is 100 m and five transmitting coils per meter are used. Therefore, it follows from this example that in a dynamic WPT system with the conventional position detection technique, 500 sensors will be required for a 100 m long path. Moreover, many wires connecting the position detection sensors to the control circuit will be necessary. Therefore, a solution to reduce the number of position detection sensors needs to be proposed and verified experimentally.

The main novelty of this paper is that we propose a solution to reduce the number of position sensors (either optical or magnetic) by a factor of two without degrading the performance of a dynamic resonant-inductive WPT system. Similar to dynamic WPT systems with the conventional position detection approach, only one transmitting coil (above or in the vicinity of which the receiving coil is located) can be activated in the dynamic WPT system with the proposed approach with the reduced number of sensors. Due to the proposed solution to reduce the number of position detection sensors (Figure 2), the overall cost of the dynamic WPT system can also be reduced. Moreover, due to the lower number of sensors and connecting wires used, the proposed solution also leads to a reduction in the level of complexity of the dynamic WPT system.

The paper is organized as follows: Section 2 presents a detailed description of the dynamic WPT system with the proposed position detection and activation approach. Section 3 is devoted to the detailed description of the designed and built scaled-down prototype of the dynamic WPT system with a reduced number of AGV position detection sensors. The most important measurement results and their analysis are also presented in this chapter. Finally, conclusions are given in Section 4.

2. Description of the Dynamic WPT System with the Proposed Position Detection Technique

Simplified schematic diagrams of the primary side of the dynamic resonant-inductive WPT system with the proposed AGV position detection approach with the reduced number of optical or magnetic sensors are depicted in Figure 2.

The magnetic position detection sensors can be either anisotropic magnetoresistive sensors or tunneling magnetoresistive sensors. The latter are more sensitive and have better response times than the former [14]. The optical sensors can be based on infra-red (IR) receivers with IR light-emitting diodes (LEDs). The price of the magnetoresistive sensors is similar to that of IR receivers with IR LEDs (according to the electronic-part search engine Octopart). The costs of the dynamic WPT systems with the two solutions for long AGV tracks will be similar because both types of sensors have similar costs. The dynamic WPT system with the optical sensors may have a slightly higher cost than the dynamic WPT system with the magnetoresistive sensors because each magnetoresistive sensor requires three connection wires, but each IR LED with an IR receiver requires four connection wires.

The control circuit is necessary in order to control the transmitting-coil-activating switches, to control the inverter, and to receive data from the sensors. In case of optical sensors, the control circuit is also necessary in order to generate pulses for IR LEDs. It can be based on a cheap (e.g., 8-bit) microcontroller (MCU) having a sufficient number of digital inputs/outputs. If it is necessary to control 16 switches from one MCU, it will be necessary to use an MCU with at least 26 digital inputs/outputs (if magnetic sensors are used for the position detection). If IR receivers are used with IR LEDs (as optical sensors) for the position detection, an MCU with a slightly higher number of digital inputs/outputs is necessary. If the inverter is based on half-bridge topology, two digital MCU outputs will be necessary for its control. The coil-activating switches can be based either on relays or on two-MOSFETs-based bidirectional switches.

As shown in Figure 2, a single inverter and a compensation capacitor may be used for a long array of the transmitting coils. However, in order to keep losses in the wires connecting the inverter and the transmitting coils low, the inverters with compensation capacitors should be placed every 3–5 m along the AGV path. If the current ratings of the compensation capacitors are not enough to withstand currents circulating in the resonant tanks, several compensation capacitors should be connected in parallel.

In order to activate a correct transmitting coil, a bar with magnets (Figure 2a) or a reflective bar (Figure 2b) should have a length equal to half the distance between the adjacent position detection sensors. As shown in Figure 3a, when the bar is above sensor 1 (near the transmitting coil L1), coil L1 is activated. When the distance between the AGV receiving coil’s center and the transmitting coil L1’s center is higher than half the distance between the centers of the coils L1 and L2, coil L1 is deactivated, but L2 is activated, because now the bar does not cover sensor 1 (see Figure 3b). If the distance between the AGV receiving coil’s center and the transmitting coil L2’s center is higher than half the distance between the centers of the coils L2 and L3, coil L2 is deactivated, but L3 is activated, because now the bar cover sensor 2 (Figure 3c). The process of the activation/deactivation of the transmitting coils repeats periodically.

In order to better understand an operating principle of the dynamic WPT system with the proposed AGV position detection approach with the reduced number of sensors, a flowchart of the AGV position detection and transmitting-coil activation is depicted in Figure 4.

The flow of the position detection and transmitting-coil-activating algorithm is as follows:

• When the first sensor (from the beginning of the track) is active, it sends logic value “1” to the MCU digital input PA4, and as a result of this, the MCU sends logic value “1” from its digital output PB10 to the driver of the first relay to energize the first transmitting coil; since other digital outputs (PB11–PB15) of the MCU yield logic value “0”, only the first transmitting coil is active.
• At the moment when the first sensor becomes inactive, the MCU digital input PA4 receives logic value “0”; therefore, the MCU sends logic “0” from its digital output PB10 to the driver of the first relay to deactivate the first transmitting coil; at the same moment, the MCU sends logic value “1” from its digital output PB11 to the driver of the second relay to energize the second transmitting coil.
• At the moment when the second sensor becomes active, the MCU digital input PB0 receives logic value “1”; therefore, the MCU sends logic “0” from its digital output PB11 to the driver of the second relay to deactivate the second transmitting coil; at the same moment, the MCU sends logic value “1” from its digital output PB12 to the driver of the third relay to energize the third transmitting coil.
• The process of the activation/deactivation of the transmitting coils repeats periodically until the moment when all the position detection sensors become inactivated (it happens when the electrical vehicle resembling an AGV has passed the last transmitting coil).
• When the electrical vehicle resembling an AGV has passed the last transmitting coil, the MCU receives logic values “0” at all its digital inputs, and therefore, all relays and corresponding transmitting coils are deactivated; the dynamic WPT system is now operating in standby mode.

The dynamic WPT system is versatile and may energize, sequentially, the transmitting coils when an AGV moves in any direction, as described in Figure 4. The flowchart (Figure 4) describes AGV position detection with either magnetic or optical sensors.

3. Experimental Part

3.1. Description of the Experimental Setup

The transmitting side of the scaled-down dynamic WPT system’s experimental prototype (with both optical and magnetic sensors) was designed according to the simplified schematic diagrams shown in Figure 2. The designed dynamic WPT system is versatile—it can be used either with connected magnetic sensors or with connected optical sensors. A microcontroller NUCLEO-F103RB evaluation board was used as the control circuit to generate complementary signals for the half-bridge inverter based on GaN system evaluation board GS66508B-EVBDB1, to control relay-based switches (via appropriate control signal converter circuit), to gather data from the sensors, and to generate 38 kHz pulses for IR LEDs (in case of dynamic WPT system with optical position sensors). The proposed position detection algorithm leading to a reduction in the number of AGV position detection sensors was implemented by using the MCU with a suitable program code created by us.

As the transmitting-coil-activating switches, factory-made 4-relay and 2-relay modules were used. As the magnetic sensors, we used anisotropic magnetoresistive sensors. As the optical position sensors, we used IR LEDs (TSAL6400) and IR receivers (TSOP38538). Schematic diagrams explaining connection of the sensors to a voltage supply and MCU are shown in Figure 5.

The receiving side of the dynamic WPT system is simple and consists of conventional components—the receiving coil, two receiving-side compensation capacitors connected in parallel, Schottky diode bridge, and two filtering capacitors (Figure 6).

The output of the designed dynamic WPT system was connected to four low-power DC motors (Adafruit 3777). The motors, through a gearbox, were mechanically connected to the wheels of the scaled-down prototype of an electrical vehicle resembling an AGV (see Figure 7). The scaled-down dynamic WPT system prototype together with the scaled-down electrical vehicle prototype was placed on a wood surface (Figure 6). Values of the main parameters of the experimental prototype are presented in

Table 1. The main parameters of the dynamic WPT system’s scaled-down prototype.

Parameter	Numerical Value	Unit of Measurement
Inductance of the transmitting coils	26	μH
Inductance of the receiving coil	26	μH
Switching frequency of the inverter	145–147	kHz
DC input voltage of the inverter	12.4	V
Tolerance on the inductances	±5	%
Total nominal capacitance of the transmitting-side compensation	2 × 22	nF
Total nominal capacitance of the receiving-side compensation	2 × 22	nF
Distance between the transmitting and the receiving coils’ ferrite pads	2.5	cm
Mutual inductance between the transmitting and the receiving coils (when they are aligned perfectly)	14	μH
Dimensions of the ferrite pad	10 × 10	cm
Distance between the centers of the transmitting coils	16	cm
AGV path length	1	m
Total number of transmitting coils	6	-

.

Since the main aim of the paper is to show that it is possible to detect the position of a moving electrical vehicle with a reduced number of sensors without degrading performance of the dynamic WPT system, we did not consider the principle of selecting the spacing between the transmitting coils. In general, the lower the spacing between the transmitting coils is, the higher the efficiency and cost of the dynamic WPT system are. Moreover, lower spacing means lower load power and voltage pulsations. We chose the distance between the centers of the transmitting coils to be equal to 16 cm because we wanted to create an AGV track with a length of ≈1 m and with the maximum number of coils equal to six.

3.2. Experimental Results and Discussion

During the experiments, the electrical vehicle moved with the average speed of ≈35 cm/s (when the switching frequency of the inverter was 147 kHz). However, when the switching frequency was 145 kHz, the electrical vehicle moved with higher speed (≈40 cm/s in average). This can also be deduced from the experimental results shown in Figure 8 and Figure 9.

Due to the fact that we did not use a voltage regulator at the secondary side, the voltage fed to the motors was not constant (see Figure 8 or Figure 9), but it was high enough to power the motors because the motors required >3 V for normal operation. When the coupling coefficient was the lowest (the receiving coil’s center was at a distance of 8 cm from the transmitting coil’s center), the input current and the output current and voltage had their maxima (at the time instants t₁ and t₅), as may be seen in Figure 8. However, at t₃ (when the receiving coil’s center was above the transmitting coil’s center), the inductive coupling was maximum and the input current was minimum (Figure 8). Therefore, at lower coupling coefficients, the motor’s rotating speed was maximum, but at higher coupling coefficients, it was minimum. These phenomena can be described by the fact that approximate output voltage of a resonant-inductive WPT system with series–series compensation is inversely proportional to the mutual inductance (and therefore, the coupling coefficient) between the transmitting coil and the receiving coil when the dynamic WPT system operates at the resonant frequency or near it [16]. Therefore, at lower inductive couplings, the output voltage of the resonant-inductive WPT system is higher. However, if the coupling coefficient is very low and a load resistance is relatively large, the output voltage may decrease as the coupling coefficient decreases. The waveforms of the output voltage, output current, and input current are quasiperiodic.

The dynamic WPT system efficiencies were measured when two different types of position detection sensors were used. The measurement results are presented in

Table 2. The measurement results of the efficiency of the dynamic WPT system with the conventional position detection method and with the proposed position detection method based on the reduced number of position detection sensors (the switching frequency was 147 kHz).

Type of the Method and Sensor Used for Position Detection	η1	η2
Proposed method, magnetoresistive sensor	31.6%	88.7%
Proposed method, optical sensor	32.7%	90.2%
Conventional method, magnetoresistive sensor	31.4%	88.3%

. The efficiency η₁ was calculated as the ratio of the system’s average output power (within time interval t₅–t₁) to the system’s average input power (within the same time interval). η₂ was calculated as the ratio of the system’s average output power (within time interval t₄–t₂) to the system’s average input power (within the same time interval). The dynamic WPT system with the optical sensors has slightly better efficiency, probably because the magnetic sensors may have had some interference due to the leakage magnetic fields. Moreover, the discrepancies may have been because the measurements results did not have high repeatability because we did not use a railroad. It should be added that the dynamic WPT system with the optical sensors operated incorrectly when there was bright light from fluorescent lamps in our laboratory. Taking into account that AGVs are traditionally used indoors (inside warehouses), we did not consider a situation when they are used outdoors (because of this, we did not consider how different weather conditions may affect the optic or magnetic sensors).

The efficiency of the dynamic WPT system was also measured when the conventional position detection approach (with a higher number of sensors, by a factor of two) was used. As can be seen in Table 2, the measured efficiency of the dynamic WPT system with the proposed solution to reduce the number of position detection sensors by a factor of two is very similar to the measured efficiency of the dynamic WPT system with the conventional position detection approach (with a larger number of position detection sensors, by two times).

4. Conclusions

Because of the proposed approach for the position detection of automated guided vehicles and transmitting-coil activation, the number of the position detection sensors is reduced by a factor of two, leading to reductions in the overall cost and level of complexity of the dynamic wireless power transfer system without degrading its performance. The proposed approach for the position detection of automated guided vehicles with a reduced number of position detection sensors may increase the popularity of the sensor-based electrical-vehicle position detection techniques in dynamic wireless power transfer systems because it can be simply implemented by using a suitable program code for a microcontroller controlling coil-activating switches and processing data from position detection sensors. To ensure the stable operation of the dynamic wireless power transfer system with magnetoresistive sensors, the negative effect of leakage magnetic fields on their operation should be minimized. If optical sensors are used for the position detection of automated guided vehicles, the negative effect of ambient light on their operation should be reduced.