Based on the sources, the use of retro-reflective materials placed in a biomotion configuration is a key strategy to improve the safety of vulnerable road users at night. Biomotion refers to the human brain's ability to perceive a moving human form from a limited set of visual cues, like lights on major moveable joints (ankles, knees, shoulders, waist, elbows, wrists).

When retro-reflective strips are positioned on these joints, vehicle headlights illuminate them, creating a pattern of moving lights that drivers easily recognize as a person. This enhances conspicuity, meaning the person "grabs" attention and is recognized as a hazard, unlike traditional reflective vests which increase visibility but might be mistaken for other objects.

Studies show significant safety advantages. Drivers recognized pedestrians with biomotion markers at distances 3.4 times greater than with a vest with the same amount of reflection. For cyclists, biomotion led to three times longer recognition distances than a vest. These benefits extend to road workers, even in cluttered environments. Biomotion marking also helps drivers perceive actions like walking direction at greater distances. The effectiveness of biomotion clothing is robust, even with factors like driver age, visual impairment, and headlight glare. This evidence influenced Australian/New Zealand standards for high visibility clothing.

Applying biomotion to recreational clothing for cyclists and runners faces challenges, as users often overestimate their own visibility. Focus groups identified key factors for user adoption: Design (appearance, practicality, preferred narrow strips), Function (comfort, convenience, integrated reflective strips), and Promotion (evidence of effectiveness, such as visibility distance statistics or demonstration videos). Ultimately, wearability and practicality are crucial for users to adopt biomotion clothing, alongside conspicuity benefits. Ongoing research aims to design and test prototypes based on this user feedback.

AI-driven pedestrian detection systems are being developed as a response to the critical global issue of pedestrian fatalities, which have risen in some regions despite overall vehicle safety improvements. These systems integrate Artificial Intelligence into vehicles to help them perceive, understand, and react to pedestrians, often more quickly than a human driver.

The core technology relies on Computer Vision to process visual data, Machine Learning and Deep Learning models (like CNNs and YOLO) trained on extensive datasets to recognize pedestrian patterns. To build a comprehensive view, Sensor Fusion combines data from multiple sensors: Cameras for visual detail, LiDAR for precise 3D mapping, Radar for range and velocity in challenging weather, and Thermal Imaging for detecting heat signatures, especially in low light. This fusion enhances object detection, classification, and helps in predicting pedestrian behavior to enable real-time decision-making, including alerts and Automatic Emergency Braking (P-AEB). Leading manufacturers and technology providers like Tesla, Volvo, Subaru, Toyota, Ford, Mobileye, Nvidia, Bosch, and Continental are implementing varying levels of these capabilities.

While statistical data indicates that P-AEB systems contribute to reducing pedestrian crashes, their real-world effectiveness is highly variable. A significant limitation is performance in low-light or nighttime conditions, where some systems have been found to be "nearly ineffective", despite most pedestrian fatalities occurring at night. Adverse weather and complex urban environments also pose challenges. Current systems can struggle with occlusion, detecting small or unpredictable targets, and are susceptible to sensor obstruction.

Key hurdles include achieving consistent accuracy and reliability across all scenarios and managing the precision-recall trade-off between avoiding false positives (unnecessary braking) and preventing catastrophic false negatives (missed detections). A major concern is algorithmic bias, with research showing lower detection accuracy for children and pedestrians with darker skin tones, often linked to biased training data. The inherent unpredictability of human behavior (e.g., jaywalking, sudden movements) also remains difficult for AI to reliably predict.

The deployment of AI also raises significant ethical and societal questions, such as determining accountability and liability in case of system failure, addressing data privacy concerns from mass surveillance, and grappling with pre-programmed ethical choices in unavoidable accidents ("trolley problem").

Regulatory bodies like NHTSA (mandating P-AEB including nighttime performance by 2029) and Euro NCAP (with influential testing protocols) are pushing for improved pedestrian safety standards. Industry standards like ISO 26262 (Functional Safety), ISO 21448 (Safety of the Intended Functionality - SOTIF for performance limitations), and the new ISO/PAS 8800 (Safety and AI) provide guidance for developing these complex systems safely. Rigorous testing involves simulation, physical tests with soft targets, and real-world data analysis.

Future trends include advanced multi-modal fusion, more accurate predictive AI, V2X communication, emerging sensors like thermal and neuromorphic cameras, and the use of Explainable AI (XAI) for transparency.

In summary, while AI has undeniably made cars "smarter" and contributed to reducing pedestrian crashes, achieving genuinely "safer streets" requires overcoming significant technological limitations, particularly in challenging environmental conditions and for vulnerable demographics, alongside vigilant ethical consideration, adaptive regulation, and building public trust.

Here is a shorter overview of pedestrian safety, drawing on the information from the sources:

Pedestrian safety is a major concern, underscored by the fact that a significant majority of fatalities occur in dark conditions, accounting for approximately 78% of pedestrian deaths according to NHTSA data. This highlights the critical challenge of reduced visibility after dark. Drivers' ability to detect pedestrians is severely diminished at night, and they may underestimate how restricted their vision becomes. This risk is exacerbated by factors like inadequate street lighting, speeding, and driver distraction.

To improve safety, enhancing pedestrian conspicuity is vital. High-visibility clothing (HVC) uses fluorescent materials for daylight visibility (especially dawn/dusk) and retroreflective materials for low-light and nighttime conditions with artificial light like headlights. Retroreflective material is crucial for nighttime visibility as it reflects light back to the source, appearing bright to drivers. Combining both provides visibility across all conditions. Standards like ANSI/ISEA 107 suggest retroreflective material should be visible from a minimum of 1000 feet. Practical advice for pedestrians includes wearing bright/fluorescent colors in the day, retroreflective materials at night, aiming for 360-degree coverage, ensuring proper fit, and using accessories that highlight motion.

Pedestrian risk is concentrated in certain areas. A striking 75% of pedestrian fatalities occur at locations that are not intersections, often happening mid-block where safe, formal crossing facilities are lacking. Uncontrolled crossings are particularly hazardous. Roadway design fundamentally impacts safety; absence of sidewalks forces pedestrians into the street, wide roads encourage higher speeds, and fatality risk rises dramatically with impact speeds above 25 mph. Parking lots and driveways also present frequent conflict points.

Key contributing factors include driver behaviors such as speeding, distraction, and impairment (alcohol/drugs involved in nearly half of driver-caused fatalities in one dataset). Pedestrian actions like distraction (phone use, earbuds) and impairment also play a role. Environmental factors like adverse weather and vehicular factors like larger SUVs can contribute. Systemic inequities, including underinvestment in infrastructure in marginalized communities, lead to disproportionately high risks for certain groups like Black and Indigenous people, older adults, and those with low socioeconomic status.

Reducing pedestrian fatalities requires a comprehensive, multi-layered approach, aligning with the Safe System Approach. Strategies include:

• Engineering: Providing sidewalks, enhancing crosswalks (e.g., RRFBs, PHBs, refuge islands, lighting, high-visibility markings), implementing traffic calming, and managing speed through design.
• Policy/Enforcement: Adopting Vision Zero/Complete Streets policies and conducting targeted enforcement of dangerous behaviors.
• Education: Reinforcing safety messages, most effective when combined with engineering and enforcement.
• Individual Responsibility: Pedestrians being visible/alert and drivers being vigilant/yielding/unimpaired.
• Community Engagement: Ensuring equitable safety planning.

Pedestrian fatalities are preventable system failures. Achieving zero fatalities demands a shift to a proactive, systemic approach that prioritizes safety for all road users.