Chinese electric vehicle giant BYD has published a new patent for an artificial intelligence-powered underbody vision system designed to detect living organisms beneath parked vehicles before the car begins to move. The technology, published by China’s National Intellectual Property Administration on June 12, 2026, under application CN122200729A, represents a novel application of computer vision aimed at preventing accidents involving children, pets, or wildlife sheltering under cars.
The system addresses a critical and largely overlooked blind spot in modern vehicle safety. While contemporary vehicles are equipped with an increasingly sophisticated array of sensors and cameras to monitor their perimeters and interiors, the area directly underneath the vehicle remains largely unmonitored. This gap poses a genuine risk when a parked vehicle begins to move, particularly in residential areas, parking lots, and other environments where children or animals may seek shelter under a stationary car.
A Two-Stage Detection Architecture
Detecting life beneath a vehicle is technically challenging due to the complex environment: shadows, varying lighting conditions, road debris, dirt accumulation, and different ground surfaces all generate noise that traditional motion detectors struggle to filter. A system that generates too many false positives—alerting the driver every time a leaf blows under the car—would quickly lose its utility.
To overcome this, BYD’s patented system employs a sophisticated two-stage baseline-image comparison method. When the vehicle is powered down and parked, the system captures an initial image of the underside and stores it as a reference, or baseline, image. This creates a customized reference map for that specific parking location and vehicle configuration.
Before the vehicle moves again, the system captures real-time images and compares them against the stored baseline. Instead of continuously analyzing the entire underbody, the AI first identifies regions that differ from the reference image. By treating static components such as suspension parts, battery enclosures, aerodynamic panels, and structural members as unchanged background, the system conserves computational resources and focuses only on newly detected objects or movement.
Once a changed region is isolated, the system applies recognition algorithms to extract feature-detection information and determine whether the new object is a living organism, such as an animal or a person. This two-stage architecture improves detection specificity while significantly reducing the false positives that would otherwise make the system impractical.
Part of a Broader Sensing Strategy
This underbody detection system is not an isolated development but part of a broader strategy by BYD to build a comprehensive vehicle sensing ecosystem. The company recently disclosed a separate patent for detecting forgotten occupants inside a vehicle using radar channel impulse response data, frequency-domain features, and angle-of-arrival analysis.
Together, these two systems address opposite sides of the vehicle: one monitors occupants left inside the cabin, while the underbody detection patent focuses on living organisms located beneath the vehicle. The strategy suggests that BYD is systematically closing the gaps in vehicle safety monitoring, combining computer vision, radar processing, and intelligent monitoring into a unified safety architecture.
This approach aligns with BYD’s broader technology roadmap, which has seen the company publish patents covering solid-state batteries, flash-charging infrastructure, and advanced driver assistance systems in recent months. While the publication of a patent application does not guarantee immediate commercial deployment—no vehicle application, launch timeline, or commercialization plan was disclosed in the document—it provides a clear indication of the direction in which BYD’s safety engineering is heading.
As humanoid robots and autonomous vehicles become more prevalent in Chinese cities, the ability to detect and respond to living organisms in complex, unstructured environments is becoming an increasingly important capability across the entire intelligent mobility sector.
BYD’s Broader AI Safety Roadmap
The underbody detection patent is one of dozens of AI-related patents that BYD has published in recent months, reflecting the company’s ambition to build a comprehensive intelligent vehicle platform rather than simply producing competitive EVs. BYD’s approach to vehicle safety is increasingly holistic: rather than adding discrete sensors to address specific risks, the company appears to be building an integrated sensing architecture that treats the vehicle as a complete perceptual system.
This strategy has significant implications for competitive positioning. As Chinese EV manufacturers compete on price and technology, the ability to offer genuinely differentiated safety features, backed by proprietary AI and sensor fusion technology, could become a key differentiator in both domestic and international markets. BYD’s aggressive patent filing activity also serves a defensive purpose, establishing intellectual property positions that could complicate competitors’ entry into specific safety technology niches as the market matures.
For consumers, the practical benefit of technologies like the underbody detection system is clear: as vehicles become more autonomous and are increasingly operated in environments where traditional driver vigilance cannot be assumed, the ability to detect and respond to hazards in the vehicle’s immediate vicinity becomes a genuine safety requirement rather than a marketing feature.