Pony.ai is moving toward broader commercialization of its robotaxi business as regulatory approvals in major Chinese cities, increased vehicle production, and expanded partnerships enable large-scale deployment of fully driverless services.
Over the past year, the autonomous driving company has secured citywide operating permits, deployed hundreds of next-generation robotaxis, and extended its manufacturing and fleet partnerships. Pony.ai has also taken steps toward accessing public capital markets. Together, these developments support expanded commercial operations in China and lay the groundwork for potential international expansion.
Citywide permits expand scope of driverless operations
In October 2025, Pony.ai was granted Shenzhen’s first citywide permit for fully driverless commercial robotaxi operations. The permit was issued jointly to Pony.ai and Shenzhen Xihu Corporation Limited, the city’s largest taxi operator, and allows driverless services across Shenzhen, with initial operations beginning in the Nanshan, Qianhai, and Baoan districts.
The Shenzhen authorization adds to Pony.ai’s existing commercial robotaxi operations in Beijing and Guangzhou. The company is currently the only autonomous driving firm approved to operate fully driverless commercial robotaxi services across all four of China’s Tier 1 cities, including Shanghai. Pony.ai has reported more than 55 million kilometers of accumulated autonomous driving mileage worldwide.
Vehicle production increases to support wider deployment
Pony.ai has also expanded its robotaxi fleet as it prepares for broader commercial rollout. Last year, the company also announced the production of its 300th ARCFOX Alpha T5 robotaxi, a vehicle jointly developed with Beijing Automotive Group.
Mass production of the Alpha T5 began in July 2025. The model is equipped with Pony.ai’s seventh-generation autonomous driving system and is designed for continuous, all-weather operation. According to the company, the Gen-7 system reduces bill-of-materials costs by approximately 70% compared with the prior generation through the use of automotive-grade components and tighter integration into the vehicle production process.
Autonomous trucking program enters pre-deployment phase
Alongside its robotaxi business, Pony.ai is expanding into autonomous freight. In November 2025, the company unveiled its fourth-generation autonomous truck lineup, developed in collaboration with manufacturing partners including SANY Truck.
The Gen-4 autonomous trucks are built entirely with automotive-grade components and are designed for a service life of up to 20,000 hours or one million kilometers. Initial deployment of battery-electric models is planned for 2026, with production expected to reach the thousand-unit scale.
Partnerships supporting manufacturing and fleet operations
Pony.ai has continued to rely on partnerships to support both manufacturing scale and capital efficiency. In January 2026, the company announced an upgraded strategic partnership with BAIC BJEV, the electric vehicle manufacturing arm of BAIC Group. The expanded agreement covers joint development of purpose-built robotaxi models, optimization of vehicle architecture and in-cabin systems, and collaboration across fleet operations and lifecycle management.
Separately, Pony.ai has adopted an asset-light expansion model through an expanded partnership with Sunlight Mobility. Under the arrangement, Sunlight Mobility funds Gen-7 robotaxi vehicles, with an initial fleet deployed in Guangzhou before the end of 2025. Fleet supply is integrated across both companies’ platforms, allowing for shared economics and reduced capital requirements.
Pony.ai’s autonomous driving technologies
In this section, we’ll explore Pony.ai’s patented technologies that address key challenges across the autonomous vehicle lifecycle, ranging from system training and sensor reliability to fleet-level passenger transport. These innovations illustrate how integrated hardware, software, and data-driven systems support safer development processes, more reliable real-world operation, and the scalable deployment of autonomous mobility solutions.
Driving emulation systems for autonomous vehicle training
Pony.ai’s patent describes a system that allows autonomous vehicles to be trained and tested in a simulated driving environment that physically moves like a real road, improving realism while avoiding the risks of on-road testing.
The problem
Training self-driving cars means exposing them to many different driving situations, such as speeding up, slowing down, turning, driving uphill or downhill, and handling rough roads. Testing all of these conditions on public roads is expensive, slow, and raises safety concerns. Because of this, developers rely heavily on simulations to train and test their systems.
Most simulations, however, exist only in software. They can show traffic and road layouts, but they do not recreate how a vehicle actually moves or feels on the road. Without this physical feedback, it can be hard to fully test how sensors and control systems work together during real driving.
How the patent solves it
U.S. Patent No. 11,391,649 describes a system that combines a virtual driving simulation with real physical movement. In this system, an autonomous vehicle is placed on a motion platform that can move the vehicle in ways that mimic real driving.
The virtual driving scene is built using navigation data and detailed maps. As the vehicle’s software responds to the simulated road such as accelerating, braking, turning, or encountering hills, the motion platform moves the vehicle to match those actions. This includes changes in direction, speed, slope, and road surface feel.
Why it matters
By connecting the virtual driving environment with real physical motion, the system creates a more realistic testing setup. Sensors and control systems receive feedback that closely matches real-world driving, without putting the vehicle on public roads.
This makes it easier to test and refine autonomous driving behavior in a safe and repeatable way. Developers can run more realistic tests, reduce development risks, and validate self-driving systems across challenging driving conditions that would be difficult or unsafe to recreate in real traffic.
The patent, titled “Driving emulation system for an autonomous vehicle,” was filed on September 5, 2019, and granted on July 19, 2022. The listed inventor is Jianan Wang.
Real-time sensor calibration for more reliable self-driving cars
The patent describes a way for self-driving vehicles to automatically keep their sensors accurately calibrated while driving, instead of relying on occasional manual checks.
The problem
Self-driving cars depend on active sensors like radar, lidar, and sonar to understand their surroundings. These sensors need to be precisely calibrated so the vehicle can correctly judge distance, speed, and direction.
Traditionally, sensor calibration is done offline in controlled environments using stationary targets. Once the vehicle is on the road, however, sensor accuracy can drift over time due to vibration, temperature changes, or slight shifts in how the sensors are mounted. When calibration degrades, it can reduce the reliability of the vehicle’s perception and decision-making systems.
How the patent solves it
U.S. Patent No. 11,454,701 describes a system that continuously recalibrates active sensors while the vehicle is driving. Instead of stopping the vehicle for calibration, the system uses Doppler data collected during normal operation.
As the vehicle moves, the sensors observe nearby objects such as road signs, curbs, trees, and other vehicles. By analyzing how these objects appear in Doppler measurements, especially those that are stationary, the system can detect small errors in sensor alignment or positioning. It then automatically adjusts sensor parameters, such as angles and mounting offsets, in real time.
Why it matters
This approach keeps sensors accurate throughout normal driving, rather than relying on occasional manual recalibration. It improves the reliability of the vehicle’s perception systems, supports safer driving decisions, and reduces downtime. Because the calibration happens automatically on the road, the method can also be applied consistently across large fleets of vehicles operating in different environments.
The patent, titled “Real-time and dynamic calibration of active sensors with angle-resolved doppler information for vehicles,” was filed on February 10, 2020, and granted on September 27, 2022. The listed inventors are Cyrus F. Abari, Harsh Mohan, and Haomin Wang.
Self-driving cars that work together to pick up passengers
The patent describes a system that allows self-driving vehicles to work together to pick up and transport passengers more efficiently in real time.
The problem
Matching passengers with autonomous vehicles quickly and reliably is a major challenge for self-driving transport services. Traditional ride-hailing or dispatch systems often rely on a central controller, manual checks, or fixed routing, which can slow response times and make it harder to manage many vehicles and passengers in busy city environments.
How the patent solves it
U.S. Patent No. 11,676,236 describes a system where multiple autonomous vehicles coordinate with each other. Instead of one vehicle handling all tasks, one car can detect a passenger who needs a ride and then alert another car that is better positioned to pick them up. Using sensors and inter-vehicle communication, the receiving vehicle navigates to the passenger, confirms they’ve entered, determines the destination, and completes the trip autonomously.
Why it matters
By sharing responsibilities across vehicles, the system makes fleet operations faster and more efficient. It reduces wait times, improves vehicle use, and allows autonomous transport services to scale in busy, complex environments.
The patent, titled “Systems and methods for autonomous passenger transport,” was filed on July 7, 2019, and granted on June 13, 2023. The listed inventors are Jialin Jiao, Jing Zhai, and Xiang Yu.
Sheppard Mullin Richter & Hampton LLP represented all featured patents.
Pony.ai: Patenting activity
Pony.ai experienced its most significant surge in patent activity in the late 2010s, reflecting an intensive phase of core technology development to support autonomous driving commercialization. Filings increased rapidly from 2017 through 2020, peaking in 2019 and 2020 at more than 120 filings per year. This period coincided with the company’s early growth following its 2016 founding, the expansion of public Robotaxi pilots in the U.S. and China, and major funding milestones, including a 2020 Series C round valuing the company at over US$5.3 billion.
From 2021 onward, patent activity moderated as Pony.ai shifted from broad platform development to operational scaling and commercialization. Filing volumes declined while the proportion of granted patents increased, indicating a focus on maturing and defending existing intellectual property. By the mid-2020s, the company appeared to rely on an established and diversified patent portfolio to support global deployment across Robotaxi, autonomous delivery, and Robotruck programs, signaling confidence in the durability of its core autonomous driving technologies.
Pony.ai: Top Law Firms
Legal representation for Pony.ai’s global patent and patent application activity between 2015 and 2025 is concentrated among a small group of highly active law firms, led by Sheppard Mullin as the most prominent advisor by volume. The firm’s substantial lead indicates a central role in managing Pony.ai’s U.S.-focused patent strategy, particularly during periods of heightened filing activity tied to core autonomous driving platform development and early commercialization.
A second tier of firms, including Zhong Lun Law Firm, Wanhuida Law Firm, and Beijing Kangxin, reflects Pony.ai’s sustained emphasis on China as a critical jurisdiction for intellectual property protection. Their involvement highlights the importance of securing domestic patent coverage alongside U.S. filings, consistent with the company’s parallel development and deployment efforts in both markets.
Additional filings handled by Bairui Intellectual Property reinforce Pony.ai’s China-focused IP strategy, while limited involvement from firms such as Gleiss Große Schrell und Partner and Chofn suggests selective engagement in Europe and other jurisdictions.
Pony.ai: Top Technology Areas
Pony.ai’s global patent portfolio is primarily concentrated in sensing, perception, and vehicle control technologies that underpin autonomous driving systems. The largest share of filings falls under G01S, covering radio direction-finding and positioning, highlighting the importance of localization, radar, and signal-based sensing for reliable navigation in complex environments.
Significant activity also appears in vehicle control and regulation, including B60W for coordinated control of vehicle sub-systems and G05D for regulating non-electric variables. These areas reflect a focus on integrated control architectures that manage steering, braking, acceleration, and motion planning, supported by additional filings under B60R related to vehicle structure and safety.
Perception and data processing are further emphasized through filings in G06V and G06T for image and video recognition and processing, as well as G06F for digital data processing. Smaller concentrations in G01C, G08G, and H04N point to supporting innovation in mapping, traffic management, and human–machine interfaces, reinforcing a comprehensive approach to autonomous mobility.
If you want to know more about the robotaxi patent landscape, you may also read our blog: https://parolaanalytics.com/blog/robotaxi-tech-patents/
