Acta Polytechnica CTU Proceedings https://doi.org/10.14311/APP.2024.51.0029 Acta Polytechnica CTU Proceedings 51:29–35, 2024 © 2024 The Author(s). Licensed under a CC-BY 4.0 licence Published by the Czech Technical University in Prague AUTONOMOUS DRIVING WITH A FOCUS ON HMI FOR PARALLEL PARKING MANEUVERS Jan Leistner∗, Zuzana Radová, Petr Bouchner Czech Technical University in Prague, Faculty of Transportation Sciences, Konviktská 20, Prague 1, 110 00, Czech Republic ∗ corresponding author: leistjan@cvut.cz Abstract. Autonomous vehicles remain a major topic in the automotive industry today, including assistance systems, cooperative communication, and other aspects that lead to full vehicle autonomy. This paper investigates vehicle autonomy with a focus on parking assistance. Along with a review of parking assistants, the paper presents a prototype parking assistant with an alternative distance display. The parking assistant prototype, compared to parking without an assistant and conventional parking assistant, was tested in a real-world experiment of a within-subjects study design. The user experience of the proposed distance representation was obtained, and the results indicated that the prototype parking assistant with an alternative distance display helps less experienced drivers get a better idea of the distance from the obstacle. Keywords: Parking assistance, in-car display, ADAS. 1. Introduction Technology is evolving rapidly in the 21st century. The manufacturing, computer, and transport sectors are particularly affected. High-speed rail lines are being built. Traffic lights are being automated, and rail safety is being improved. The car industry is also moving forward with the development of autonomous vehicles. An autonomous vehicle can independently assess and respond to traffic scenarios without driver intervention. The Society of Automotive Engineers (SAE) has categorized vehicle autonomy into six levels (0–5), with increasing reliance on Advanced Driver Assistance Systems (ADAS) that must be fully reli- able and cooperative [1]. The extent to which driver assistance systems are implemented varies depending on the level of autonomy. These consist of driving aids that maintain speed or keep the vehicle within the lane and assistants that aid in parking or ensure safe exit from the vehicle for passengers. Parking aids with a numerical distance display are widely avail- able in the aftermarket. Therefore, an experiment was designed to test whether the alternative obstacle distance display is a suitable aid for the driver and can be used to park the vehicle closer to the obstacle than using the distance display with a color scale and audio signals. Tesla’s numeric obstacle distance dis- play is currently used with both the ultrasonic sensor and vision systems. The vision system replaces the ultrasonic sensors with cameras. The autonomy of vehicles dates back to the early part of the last century [2]. The article says the first autonomous vehicle, controlled by a radio controller, traveled through the streets of New York City in 1925. As mentioned in [3], the subsequent development in this field was the autonomous vehicle built in 1956, guided by electromagnetic circuits embedded into the roadway. However, using it on the streets was too expensive, and the technology was limited to airport roads. An autonomous vehicle that required no modi- fied infrastructure was introduced in 1995 [4, 5]. The test route was from Munich to Odense in Denmark, and the vehicle was in autonomous mode for 95% of the journey. Following the introduction of GPS navigation systems and digital maps in 2000, there was a significant enhancement in research into vehi- cle autonomy, with technology giants such as Google beginning to develop autonomous vehicles. 2. Autonomous vehicles Autonomous vehicles utilize sensors and cameras to collect real-time information about their surround- ings [6]. The data is then evaluated by an on-board computer, which works alongside map data and in- teracts with surrounding vehicles and infrastructure. This system enables the vehicle to navigate roads and be conscious of potential obstructions within its immediate environment. The ultimate aim of this technology is to achieve autonomous operation with- out driver’s intervention. However, attaining this objective demands prolonged effort and extensive re- search. Vehicle autonomy is distinguished into six lev- els. At present, autonomous vehicle development lies at Level 3. Google showcased a Level 3 autonomous vehicle capable of autonomously driving in specific suburban regions [7]. In the Czech Republic, cars with Level 2 autonomy can be purchased, offering a ’hands- off’ mode in which the vehicle handles steering wheel rotation, acceleration, and braking. This category of autonomy encompasses Tesla’s Autopilot [8] and Cadillac’s Super Cruise system [9], amongst others. However, it is worth noting that the driver must pay full attention to the road. 29 https://doi.org/10.14311/APP.2024.51.0029 https://creativecommons.org/licenses/by/4.0/ https://www.cvut.cz/en J. Leistner, Z. Radová, P. Bouchner Acta Polytechnica CTU Proceedings Figure 1. Levels of autonomy [1]. According to SAE, the six autonomous driving lev- els can be divided into two groups (see Figure 1). Levels 0–2 are classified as driver-support systems, while levels 3–5 are classified as automated driving. 3. Parking assistance Parking assistance has become a popular topic. As the number of vehicles on the road increases, parking areas reach their limit. Drivers often struggle to park their vehicles in designated spaces due to the narrow parking spaces. Therefore, vehicle manufacturers are focusing on the development of parking assistance, among other things. Since 1933, car manufacturers have enhanced driver convenience in parking [10]. During that year, a car with a special wheel perpendicular to the vehicle’s longitudinal axis was introduced to park effortlessly. Another notable innovation was Volkswagen’s Futura model [11], which enabled all four wheels to pivot at a right angle, resulting in comfortable parking. In 2003, Toyota introduced the Prius equipped with the Intelligent Parking Assist system [12], shown in Fig- ure 2. The car featured sonar sensors and cameras on both of its bumpers. When drivers decided to park the vehicle, they selected the appropriate park- ing space on the radio display. Once a space was selected, the car’s system took control of the vehicle and parked it in the chosen space. Initially, there were issues concerning the recognition of pedestrians and Figure 2. Intelligent Parking Assist in Prius [13]. particular objects such as a pram. Toyota was the only manufacturer to offer this system for three years until other manufacturers, including BMW and Ford, adopted the system in 2006. 4. Ultrasonic sensors The Prius’s sonar sensors were replaced by ultrasonic sensors. This newer approach offers more sophistica- tion regarding its integration into the vehicle’s struc- ture and ease of operation. Ultrasound is a mechanical wave with a frequency greater than 20 kHz [14]. The speed of ultrasound propagation relies on the density of the medium through which the waves travel. The 30 vol. 51/2024 Autonomous driving with a focus on HMI for parallel parking maneuvers velocity of ultrasound is the greatest in solids, reaching up to 3 300m s−1. Conversely, ultrasonic waves spread the slowest in gases at 345.7m s−1. The propagation speed in liquids approximates 1482m s−1 [15]. The speed of ultrasonic waves is contingent upon the density and bulk modulus of the medium they traverse. This relationship is well established and expressed by Equation (1). v = √ K ρ , (1) where K is the bulk modulus and ρ is the density of the medium. Ultrasonic sensors work based on calculating the speed at which the signal bounces off an obstacle [16]. The sensor consists of a transmitter and a receiver. The distance is determined by the transmitter sending an audio signal and triggering a time measurement. The transmitted signal propagates through the envi- ronment until it encounters an obstacle from which it bounces. The reflected signal then travels back to the sensor, which listens for the “echo”. After the transmitted signal returns to the sensor, a distance cal- culation is performed. The sound propagates through the air at approximately 340m s−1, which means that a one-centimeter wave travels in 2.94 × 10−5 seconds. Equation (2) shows the formula for calculating the distance. d = t · c 2 , (2) where d is the calculated distance, t is the time taken for the ultrasound to return to the sensor, and c is the speed of sound propagation. The variable c depends on the environment in which the sound propagates. The receiver is equipped with a piezoceramic sensor that can change its geometric shape based on the electrical signal, so it can be said that the piezoceramic converts mechanical energy into electrical energy [17]. 4.1. Usage of ultrasonic sensors in parking assistance 4.1.1. BOSCH Automated Parking System Sensors based on ultrasonic waves work in conjunction with a computer that analyses the proximity of objects in the vicinity of the vehicle. BOSCH states [18] that sensors are mounted on the sides of the vehicle to scan and map possible parking spaces. Once the assistant has confirmed that the space it has scanned is suitable for the car to park in, it sends a signal to the driver. Upon pressing a button, the assistant calculates the optimal route to the parking space, determining the required steering wheel turns and necessary maneuvers. Once the path is determined, the driver can relinquish control and guide the vehicle with gentle pressure on the accelerator and brake pedals. The assistant uses the electronic steering wheel control to guide the vehicle to the parking spot. Figure 3. Parking scenario’s scheme. The driver remains in full control of the maneuvers and may cancel the operation anytime by touching the steering wheel. In addition, the assistant supervises the exit from the parking space and brings the vehicle to a safe position for the driver to proceed to the traffic. 4.1.2. Valeo Automatic Parking Valeo introduced the function Park4U in 2007 [19], which was implemented in the Volkswagen Touran. The parking maneuver is similar to the BOSCH sys- tem in the hands-free mode. Valeo states that once ultrasonic sensors have located a fitting parking space, the driver only needs to shift into the reverse gear, and the system takes control of the vehicle. The driver can interrupt the system’s operation by touching the steering wheel. Additionally, in 2011, Valeo launched the Park4U Remote mobile application, which allows drivers to leave their vehicles and park them using their phones. This distinguishes it from the BOSCH parking assistant as it eliminates the need for the driver to remain inside the car to operate the brake and accelerator pedals. 5. Experiment An experiment was designed to test the user experi- ence using different HMI elements in parking assis- tants. Parking was done on a remote road with no surrounding traffic. The experiment aimed to deter- mine if and which parking assistance could better help drivers while parking. A Suzuki Ignis was used for parking without a parking aid and with the proto- type, and a Skoda Yeti was used for parking with a conventional parking assistant. 5.1. Parking space For the purpose of the experiment, a parking space was created. The parking space was designed to conform to the dimensions specified in standard CSN 73 6056 [20]. The length of the parking space was 5 750mm, and the depth was 2 000mm, as illustrated in Figure 3, and Figure 4 shows the testing. 5.2. First parking Parking without parking assistance. Drivers were asked to park the vehicle using only the rear-view mirrors. 31 J. Leistner, Z. Radová, P. Bouchner Acta Polytechnica CTU Proceedings Figure 4. Parking scenario in real testing. Figure 5. HMI in Skoda Yeti vehicle. 5.3. Second parking The second method was parking the vehicle with the help of a parking assistant fitted to the car by the manufacturer. A 4-channel ultrasonic sensor system with a display of the distance to the obstacle on the radio display, paired with acoustic signals, was used. Human-Machine Interface, shown in Figure 5, informs the driver about the distance from the obstacle. 5.4. Third parking Parking using the prototype parking assistant. The parking assistant, equipped with an ultrasonic sensor, responds to approaching an obstacle with an acoustic signal and displays the obstacle distance using col- ored LE Diodes and an OLED display that shows the distance in centimeters. 5.5. Implementation of prototype design Before building the prototype parking assistant, a sim- ulation of the circuit, shown in Figure 6, was per- formed in a virtual environment. The parking assis- tant consists of an ARDUINO microcomputer, acting as a control element. ARDUINO processes data from the ultrasonic sensor and calculates a distance from the obstacle. Based on the measured distance, partic- ular LEDs are powered up. The connection of the LE Diodes requires a combination with a corresponding resistor. The value of the required resistance for each Figure 6. Virtual simulation of the prototype circuit. diode was determined using Ohm’s law expressed by Equation (3). The microcomputer supplies the diode with a voltage of up to 5V, and the current required for proper diode operation is 25mA. R = U I = 5 0.025 = 200 Ω (3) The circuit also includes a piezoelectric buzzer fitted with a 100 Ω resistor to reduce the volume. The LEDs were wired to connect a selected 200 Ω resistor to the cathode. The positive pole of the LED is connected to the digital input of the ARDUINO UNO board. All eight LEDs were connected to pins 6 to 12, leaving the other pins available for other components. The piezoelectric buzzer was connected with a 100 Ω resistor on the positive pole to the analog input A0 and the negative pole to the GND pin on the breadboard. The TRIG and ECHO pins of the ultrasonic sensor were connected to digital pins 2 and 3, respectively. The positive and negative poles of the component were connected to the breadboard, which was connected to the ARDUINO. In the next stage, an OLED display was added to the circuit to show the distance in centimeters measured by the ultrasonic sensor. The OLED display has four pins, which were connected as follows: the VDD pin was connected to the positive pole of the ARDUINO, the GND pin to the negative pole, the SCL pin to the analog pin A5, and the SDA pin to the analog pin A4 on the ARDUINO board as shown in virtual simulation in Figure 6. The source code was created to ensure the correct functionality of all involved components. The dis- tances at which the LE diodes light up and the buzzer sounds were defined in the code. The measured dis- tance is also displayed in centimeter (cm) units on the OLED display. The first detection distance was set to 150 cm based on data collected from vehicles that had the parking assistant fitted by the manufacturer. The smallest permissible distance, so-called Full Tone, was set to 30 cm. All the LEDs light up at the full tone 32 vol. 51/2024 Autonomous driving with a focus on HMI for parallel parking maneuvers distance, and the buzzer provides an uninterrupted tone. 6. Results In the experiment, the participants were asked to park the vehicle in all three ways. The time needed to park the car and the final distance from the obstacle were measured. The experiment was performed with 11 participants of different age groups and driving experience. Participants were also asked whether their vehicle had been equipped with parking assistance of any kind, see Table 1. ID Age Km/Year Parking Assistance 1 20 5000 Yes 2 20 1500 No 3 20 2500 Yes 4 21 600 No 5 20 3500 No 6 19 1500 Yes 7 50 9000 No 8 52 9000 No 9 47 38500 Yes 10 50 16500 Yes 11 53 21000 No Table 1. Data table. ID Time Distance 1 00:20,1 146 2 00:48,0 115 3 00:43,2 119 4 02:21,2 107 5 00:36,2 150 6 00:51,8 129 7 00:15,0 129 8 00:28,5 141 9 00:48,0 103.5 10 00:40,0 89 11 01:20,0 170 Mean Values 00:47,2 127.14 Table 2. Data without parking aid. The results of parking without a parking assistant are shown in Table 2, and the results of parking with the conventional parking assistant are demonstrated in Table 3, indicating that the mean distance decreased by 33.23 centimeters compared to parking without a parking aid. However, the mean time needed for parking increased by 8.9 seconds. The data collected shows that the mean value of the final distance from the obstacle gradually decreased with the use of parking assistance. In the third part of the experiment, the mean distance decreased even more, as seen in Table 4. The mean time needed for parking changed marginally compared to parking ID Time Distance 1 00:19,9 77 2 01:24,7 110 3 00:40,8 76 4 01:18,9 68 5 01:16,4 106 6 01:45,5 113 7 00:18,0 110 8 00:31,6 134 9 01:00,1 54 10 00:40,7 89 11 01:00,1 96 Mean Values 00:56,1 93.91 Table 3. Data with conventional parking assistant. ID Time Distance 1 00:22,1 22 2 00:35,8 51 3 00:44,9 91 4 01:21,8 150 5 00:46,7 76 6 01:36,0 145 7 00:14,7 116 8 01:02,0 121 9 00:54,7 57 10 00:24,3 93 11 00:36,8 67 Mean Values 00:47,3 89.91 Table 4. Data with prototype parking assistant. with a vehicle without a parking assistant. Still, com- pared to the mean time of parking with a conventional parking assistant, the mean time for parking with the prototype decreased by 8.8 seconds. Using the prototype, the participants parked an average of 37.23 centimeters closer to the obstacle than using a car without parking assistance. This might have occurred because the drivers were comfortable with the numerical display of the distance. Driver ID 6, despite having a parking aid in his car, was able to park faster using only the rear-view mirrors. The reason for this may be that the parking aid can also be stressful for some drivers and can affect the driver’s concentration when parking. However, the mean values of the final distance show that the drivers were comfortable with the alternative distance display, resulting in a close approach. They were also able to park the vehicle more quickly than when using conventional parking assistance. Figures 7 and 8 show the measured data during the experiment. In Figure 7, a greater parking distance from the obstacle can be observed for most partici- pants when using rear-view mirrors. The distance was measured from the rear bumper of the parked vehicle to the nearest part of the obstacle to which the vehicle 33 J. Leistner, Z. Radová, P. Bouchner Acta Polytechnica CTU Proceedings Figure 7. Comparison of distances from the obstacle. Figure 8. Comparison of the parking times. was parked (a third-party parked vehicle). Figure 8 shows the time it took to park the car. The time was measured from the start of the parking maneu- ver to its completion when participants applied the parking brake. For participants ID 5 and 6, a notable increase in parking time can be observed, which can be explained by trying to park the vehicle as close as possible to the obstacle using the parking assistant. Therefore, the measured data show that both conven- tional and prototype parking assistance improve the driver’s awareness of the proximity of the obstacle but increase the time required to park the vehicle. The parking aid prototype with an alternative distance display is better used on roads with less traffic. 7. Conclusion This paper aimed to present the prototype parking assistant and verify whether the alternative distance display could be a suitable aid for drivers. It was explored that the assistant could help park the vehicle closer to the obstacle than when using the conven- tional parking assistant. The experimental results showed that the parking assistant with the alternative distance display gave drivers a better idea of the dis- tance from the obstacle. Therefore, the data mean of the final distance to the obstacle was smaller than that of the other parking aids. The mean parking time was also the shortest when using the alternative distance display. For more experienced drivers, the numerical representation of distance from obstacles made a less significant difference than for those with less expe- rience. The decreased time spent using the parking aids is also more evident for the less experienced and younger participants. The authors believe that having a larger sample of participants would provide a more comprehensive analysis. This should be addressed in the forthcoming experiments. Furthermore, future research on park- ing assistants might also focus on analyzing drivers’ subjective impressions of the user environment, semi- automatic and automatic parking assistants, and the strategy the system uses to park the vehicle in the parking spot. References [1] Levels of driving automation, 2021. [2021-08-04]. https://www.sae.org/binaries/content/gallery/ cm/content/news/sae-blog/j3016graphic_2021.png [2] History of the autonomous car, 2012-2021. [2021-08- 05]. https://www.titlemax.com/resources/history- of-the-autonomous-car/ [3] Science: Radio auto. [2021-08-06]. http://content.time.com/time/subscriber/ article/0,33009,720720,00.html [4] J. Janai, F. Güney, A. Behl, A. Geiger. Computer Vision for Autonomous Vehicles. Foundations and Trends, Boston, 1st edn., 2020. https://doi.org/10.1561/0600000079 [5] M. Maurer, R. Behringer, S. Furst, et al. A compact vision system for road vehicle guidance. In Proceedings of 13th International Conference on Pattern Recognition, vol. 3, pp. 313–317. IEEE, 1996. https://doi.org/10.1109/ICPR.1996.546962 [6] The 6 levels of vehicle autonomy explained. [2021-08-05]. https://www.synopsys.com/automotive/autonomous- driving-levels.html [7] WAYMO. Waymo driver, 2023. [2024-03-15]. https://waymo.com/waymo-driver/ [8] Tesla. Tesla Autopilot, 2024. [2024-03-15]. https://www.tesla.com/cs_CZ/autopilot [9] Cadillac. Cadillac’s Super Cruise, 2021. [2024-03-15]. https://www.cadillac.com/technology/super- cruise [10] The ingenious “fifth wheel” parallel parking tool that never hit it big, 2018. [2024-03-15]. https://www.hagerty.com/media/automotive- history/parallel-parking-made-easy-1930s- style/ [11] Volkswagen futura concept 1992, 2019. [2021-08-06]. https://www.autorevue.cz/getthumbnail.aspx?w= 1600&h=1000&q=60&id_file=266246351 [12] Toyota unveils car that parks itself, 2003. [2023-10-27]. http://edition.cnn.com [13] Intelligent Parking Assist System, 2001. [2021-08-05]. https://en.wikipedia.org/wiki/Intelligent_ Parking_Assist_System 34 https://www.sae.org/binaries/content/gallery/cm/content/news/sae-blog/j3016graphic_2021.png https://www.sae.org/binaries/content/gallery/cm/content/news/sae-blog/j3016graphic_2021.png https://www.titlemax.com/resources/history-of-the-autonomous-car/ https://www.titlemax.com/resources/history-of-the-autonomous-car/ http://content.time.com/time/subscriber/article/0,33009,720720,00.html http://content.time.com/time/subscriber/article/0,33009,720720,00.html https://doi.org/10.1561/0600000079 https://doi.org/10.1109/ICPR.1996.546962 https://www.synopsys.com/automotive/autonomous-driving-levels.html https://www.synopsys.com/automotive/autonomous-driving-levels.html https://waymo.com/waymo-driver/ https://www.tesla.com/cs_CZ/autopilot https://www.cadillac.com/technology/super-cruise https://www.cadillac.com/technology/super-cruise https://www.hagerty.com/media/automotive-history/parallel-parking-made-easy-1930s-style/ https://www.hagerty.com/media/automotive-history/parallel-parking-made-easy-1930s-style/ https://www.hagerty.com/media/automotive-history/parallel-parking-made-easy-1930s-style/ https://www.autorevue.cz/getthumbnail.aspx?w=1600&h=1000&q=60&id_file=266246351 https://www.autorevue.cz/getthumbnail.aspx?w=1600&h=1000&q=60&id_file=266246351 http://edition.cnn.com https://en.wikipedia.org/wiki/Intelligent_Parking_Assist_System https://en.wikipedia.org/wiki/Intelligent_Parking_Assist_System vol. 51/2024 Autonomous driving with a focus on HMI for parallel parking maneuvers [14] I. G. Main. Kmity a vlny ve fyzice [In Czech; Oscillations and Waves in Physics]. Academia, Praha, 1st edn., 1990. [15] Ultrazvukové vlnění [In Czech; Ultrasound Waves], 2020. [2024-03-16]. https: //www.wikiskripta.eu/w/Ultrazvukové_vlnění [16] Z. Malá, D. Nováková, T. Vítů. Fyzika I [In Czech; Physics I]. České vysoké učení technické v Praze, 3rd edn., 2009. [17] Ultrazvukové senzory přiblížení – funkce, provedení, použití [In Czech; Ultrasonic proximity sensors], 2017. [2021-08-10]. https: //automatizace.hw.cz/ultrazvukove-senzory- priblizeni-funkce-provedeni-pouziti.html [18] Park assist. [2024-03-14]. https://www.bosch-mobility-solutions.com/en/ solutions/parking/parking-assist/ [19] Valeo Park4U® an automated parking system to park easily, 2021. [2021-08-05]. https: //www.valeo.com/en/park4u-automated-parking/ [20] ČSN 73 6056 Odstavné a parkovací plochy silničních vozidel [In Czech; Parking areas for road vehicles], 2011. Úřad pro technickou normalizaci, metrologii a státní zkušebnictví. 35 https://www.wikiskripta.eu/w/Ultrazvukov�_vln�n� https://www.wikiskripta.eu/w/Ultrazvukov�_vln�n� https://automatizace.hw.cz/ultrazvukove-senzory-priblizeni-funkce-provedeni-pouziti.html https://automatizace.hw.cz/ultrazvukove-senzory-priblizeni-funkce-provedeni-pouziti.html https://automatizace.hw.cz/ultrazvukove-senzory-priblizeni-funkce-provedeni-pouziti.html https://www.bosch-mobility-solutions.com/en/solutions/parking/parking-assist/ https://www.bosch-mobility-solutions.com/en/solutions/parking/parking-assist/ https://www.valeo.com/en/park4u-automated-parking/ https://www.valeo.com/en/park4u-automated-parking/ Acta Polytechnica CTU Proceedings 51:29–35, 2024 1 Introduction 2 Autonomous vehicles 3 Parking assistance 4 Ultrasonic sensors 4.1 Usage of ultrasonic sensors in parking assistance 4.1.1 BOSCH Automated Parking System 4.1.2 Valeo Automatic Parking 5 Experiment 5.1 Parking space 5.2 First parking 5.3 Second parking 5.4 Third parking 5.5 Implementation of prototype design 6 Results 7 Conclusion References