Abstract

In pedestrian protection, there are multiple technologies available along with passive and active methods to avoid bus-to-person collisions. The cost of the technology and effectiveness are not always relative nor is complexity beneficial. In this paper, it is proposed that physical run-over mitigation is more effective than passive detection or active vehicle intervention mitigation. [Radar and other devices alone are not sufficient to preventmitigate injury or death—a physical barrier is required.]  Safety technology, such as the S-1 Gard and Smart Barrier-S (covering the side of the bus) present a last line of defense. ] The proposed framework consists of the analysis of available technology that can be readily fielded along with the classification of the detection and mitigation. Results are based on historic cases recorded by organizations such as the US NHTSA and experimental methods demonstrating suitable solutions in academic settings. 

Introduction

Run-over accidents are one of the most exigent issues in public transit safety. The Vision Zero movement highlights the urgency of mitigating physical injury and death. The results of a run-over accident are nearly always fatal or with severe injury. [Due to the weight of the bus itself, including the heavy frame and engine, persons coming in contact with the bus are at greater risk than with light-duty trucks and common passenger vehicles.] Despite advancements in technology and equipment, there has not been a significant reduction in incidents between buses and vulnerable road users (VRU). Meanwhile, driver proficiency has been on the decline due to a shortage of qualified drivers while the experienced operator pool is retiring out of the workforce. In addition to distractions to the operator, and vulnerable road users, the results are clear that there are more opportunities for bus-to-person collisions. 

According to the US NHTSA, children and cyclists are typically amongst the highest percentage of incidents with severe injury or fatalities. Further, scooters, such as Lime and Bird, along with skateboards are presenting a growing hazard for bus operators. The current technology is specifically geared towards the mitigation of incidents with these two groups due to the volume and severity of incidents. Regulation ranges from a state to international level to specifically call for technology to warn and prevent collisions.  [On September 19, 2023, FTA issued Safety Advisory 23-1: Bus-to-Person Collisions to recommend thato transit agencies that provide bus service consider mitigation strategies to reduce bus-to-person collisions to help reduce the likelihood and severity of bus collisions with pedestrians, bicyclists, and micromobility users.]  

The National Safety Council in 2021, reports that 22% of pedestrian collisions occur at intersections and nearly 84% in urban settings. Regulations such as the UN ECE R151 and GB 11567 make sense as a step in the right direction toward reducing incidents. The UN ECE R151 requires for detection and warning of VRU in a specified zone to the side of the vehicle. China’s GB 115672 requires vehicles to have lateral underrun (run-over) protection requirements, a physical barrier requirement published in 2017. 

According to NHTSA, the most vulnerable group of pedestrians, ages 60-70 made up 23% of incidents with fatalities. This age group makes up less of the walking population, so the statistics are disproportionate due to many contributing factors. These factors include fragility associated with the aging process, difficulty crossing the curb, slower walking speeds, and difficulty with judgment and interactions at intersections. The elderly have very little chance of survival if they are struck and/or run over by the vehicle.  [Further, scooters, such as Lime and Bird, along with skateboards and bicycles are presenting a growing hazard for bus operators.]

Methods of Detection

The most classic and prevalent type of aid to detect VRU is by direct observation, which is via the operator’s eyes through windows and mirrors. As a basic requirement in automobiles, including standards such as the US 49 CFR 393.80, there are exterior mounted mirrors used to view down the sides of the vehicle. Additional mirror types such as cross-view mirrors further aid in the operator detecting VRU in blind spots due to the vehicle’s shape and size. This method is the most robust as there are no electronics to fail, they are low-cost and easy to maintain. As expected, there are a limited number of directions that an operator can observe at a single time being the ultimate limitation. In addition, the number of distractions increases with traffic and technology causing a further reduction in active and meaningful observations through the mirrors. 

Cameras are now mainstream and are specified in regulations such as the previously discussed US 49 CFR, specifically under section 571.111. These systems called are referred to as CMS, Camera Monitor Systems, are also regulated under ISO 16505 to perform effectively. The German BASt published a report which stated that a CMS could provide sufficient visibility for drivers. With properly employed camera systems and display products, cameras can be an improvement to mirrors alone. Rearward visibility is the primary requirement with the current legal requirements in most locales. There are limited studies and applications of CMS being the sole technology for side visibility as of today. Additional technology with AI (artificial intelligence intelligence) or MOD (moving object detection) could aid in the detection of VRU in camera-based systems. The obvious deficiency of camera-based systems is that environmental conditions and installation points are critical along with frequent upkeep. 

Radar or electronic detection is now amongst one of the fastest growing and widely adopted advanced technologies in blind spot detection including VRU and for vehicular traffic. Systems typically use microwave or millimeter wave frequencies as they are commercially available and reasonably priced for the amount of precision and accuracy that can be delivered. Requirements such as the UN ECE R151, GB/T 39265-2020 and Transit for London, Progressive Safety System: Detailed Specifications for Blind Spot Information System have clear specifications for coverage areas which that are practical and relevant to protect against run-over incidents. Radar systems have enough reliability to provide input to have autonomous mitigation to command behavior on the vehicle in addition to passive warning systems. Being that the systems utilize radio waves to obtain their data, it is very resistant to weather and lighting conditions unlike camera systems making it a stronger candidate for detection technology. 

LiDAR is one of the newest methods of object detection capable of pedestrian detection. LiDAR (light detection and ranging) utilizes a laser to send a pulse and then measure measures the time for the reflected light to return to the sensor, determining the range. Because of the nature of the beam pulse, you can have a 3D image constructed from the data providing a very accurate and precise image. With the new solutions for LiDAR, this technology has become much more accessible in the past few years with very small units. Although there hasn’t been wide adoption of the technology yet, there should be a vast array of choices within the near future. Based on the current standards required for blind spot detection with VRU protection, there is no reason why LiDAR cannot be employed as a solution alternative to existing technology. The LiDAR sensors have a similar limitation to camera systems where the lenses are subjected to environmental conditions that reduce performance similar to a camera system, conditions of fog and rain can severely impact the data quality. 

Methods of Mitigation

The lowest level of mitigation is by way of warnings exterior of the vehicle. This is covered by the use ofusing signage or electronic warnings such as lights and audible warnings. Although this does not provide any intelligence to the driver, it can raise awareness of the VRU as specified in the FORS (Operator Recognition Scheme) standard. There is no judgment or driver interaction other than to do their equipment inspectionsinspect their equipment to ensure all features are working as intended. The FORS standard is a system of tiered levels of accreditation in the United Kingdom. As the level of safety measures increases, the operator will qualify for higher accreditation levels. 

Most detection systems provide the operator a warning by audible, visual, and haptic feedback [the use of advanced vibration patterns to convey danger]. In some cases, there is also an exterior warning which can be audible or visual as well to prevent contact between the VRU and the vehicle. Audible warnings are currently available in the form of an alarm sounding a series of beeps, sirens, or voice prompts to gain the attention of the operator and/ or VRU. Visual warnings are delivered in the form of indicator lamps, a display mechanism, integrated messages on the gauge cluster or infotainment display, or exterior illumination. The visual warnings are to provide information to the operator or to gain the attention of the operator and/ or VRU. Haptic feedback is available only to the driver as they have an assigned seat in the vehicle. The haptic feedback is a vibration unit that is either installed to vibrate the steering wheel assembly or to provide vibration to the driver at the seat back or bottom. These warnings satisfy the requirements of UN ECE R151 and GB/T 39265-2020. All of these display methods require the driver to pay attention to monitor those warnings, process the information, and then decide what to do with the information to deviate from their behavior in controlling the vehicle. If the judgment of the operator is not sufficient or appropriate, regardless of how effective the detection and warning method is, the results could still be catastrophic. 

The next level of incident mitigation is active mitigation. However, this particular method of mitigation is better characterized as reactive mitigation. A system must first be able to detect the potential incident, and then have the correct logic to decide to output a behavior. In most instances, this behavior is stopping the vehicle. AEB (Autonomous Emergency Braking) is prevalent for forward collision mitigation and parking assistance, but is relatively new for VRU protection. The level of intrusiveness requires careful adjustment in order toto not apply the brakes when not desired. In the application of transit buses, accidentally full locking of the brakes by the VRU system could result in rider injuries due to the sudden change of speed. However, reactive mitigation removes the need for the driver to maintain attention to a particular area to make a decision improving the likelihood of mitigating an incident.  

Lastly, a truly active mitigation shall be characterized by a physical barrier to prevent the VRU from being in a position to be run over. Standards such as the GB 11567-2017 outline the requirement for a physical barrier to be used in heavy vehicles to physically block a VRU from being run over. By providing a physical barrier, the VRU would be pushed out of harm’s way not allowing for them to be put in the wheel’s travel path. These physical barriers can be fashioned in a number of different products, they can be metal guards on a high chassis, devices on the body and chassis which deflect the VRU if they are on the ground in the proximity of the tire, or they can be actively deployed barriers or deflectors using electronics or pneumatics which allow retracting to allow for more road clearance when there is no danger.   [Radar is most effective when loading and unloading passengers, however, once a bus is in motion, radar becomes ineffective to prevent injury or death. ]  An active mitigation device requires no interaction or decision-making process from the driver, driver or VRU distraction does not change the efficacy of the system. Due to the wide range of effective barrier systems, the cost of this protection method could be very economical making it quite practical.  [Public Transportation Safety Int’l Corp.’s long standinglong-standing physical barrier, known as the “S-1 Gard” and more recently developed technology implementing both a radar and physical barrier solution, known as the “Smart Barrier-S”, are examples of barrier systems which are both effective without being cost prohibitiveand economical.]

Conclusion

There are run-over accidents worldwide, the results nearly always being fatal or including serious injury. Globally, there are regulations, recommendations, and organizations that have made compelling cases to prevent incidents by employing technology, practices, and solutions. These various organizations have similar recommendations and standards allowing for fleets, manufacturers, and retrofitters to offer universally compliant solutions. Attention and decision making is the highest risk to incident mitigation, regardless of the operator or VRU. Active barrier mitigation is likely the most effective mitigation method as it provides a final stop to the run-over incident. A physical barrier such as the S-1 Gard is an effective example that can mitigate a run-over accident.  In conjunction with an effective detection method and notification solution, the technology exists to improve awareness and advanced warning before the impact. However, warnings utilizing cameras, radar, and LiDAR alone are not enough. 

About the Author

Augustin Leung is an industry veteran with over three decades of transportation experience and over a decade specifically focused in on safety. Born and educated in California, his career began in the aftermarket automotive sector then advanced into manufacturing in high-technology fields with prestigious companies such as Clarion, AAMP Global, and CUB Elecparts. Augustin has served large-scale customers enhancing vehicle safety including Thomas Built Buses, Lippert Components, and Ferrari. Augustin has held executive positions and advisory roles for passenger and pedestrian safety committees. 

References

UN ECE R151 – United Nations, Uniform provisions concerning the approval of motor vehicles with regard to the Blind Spot Information System for the Detection of Bicycles (November 2019)

GB 11567-2017 – National Standard of the People’s Republic of China, Motor vehicles and trailers – lateral and rear underrun protection requirements (September 2017)

GB/T 39265-2020 – National Standard of the People’s Republic of China, Road vehicles – Performance requirements and testing methods for blind spot detection (BSD) system (June 2021)

TFL PSS: Technical Specifications (BSIS) (September 2023)

ISO 26262-1:2018 – Road vehicles – Functional safety (December 2018)

ISO 16505: 2019 – Road vehicles – Ergonomic and performance aspects of Camera Monitor Systems (July 2019)

Comparison of Pedestrian Detectors for LiDAR Sensor Trained on Custom Synthetic, Real and Mixed Datasets; AGH University of Science and Technology Cracow, Poland, Lodz University of Technology Lodz, Poland (September 2022)

Vision based pedestrian detection for advanced driver assistance; Savitribai Phule Pune University Maharashtra, India (April 2015)

FMCSA (Federal Motor Carrier Safety Administration) DOT Tip Sheet (August 2019)

DOT RITA UTC Spotlight May 2013: The Potential of Side Cameras to Reduce Bus Side Collisions

FMVSS 111 – US DOT NHTSA Federal Motor Vehicle Safety Standard: Rear Visibility (May 2018) 49 CFR 571.111

49 CFR 393.80 – US Code of Federal Regulations Title 49, Subtitle B, Chapter 3, Subchapter B, Part 393, Subpart G, section 393.80: Rear-vision mirrors

EN12889-1:2007 – Fixed, vertical road traffic signs