StradVision, an artificial intelligence-based automotive object recognition software company, will complete the first phase of development for "MultiVision," a system targeting Level 3 and Level 4 autonomous driving, by next year. The company will also commercialize a road data processing platform built on its partnership with Amazon Web Services as a separate software-as-a-service product.
At the autonomous driving test center inside StradVision's Dongtan office in Hwaseong, Gyeonggi Province, on the 8th, two passenger cars stood out — one with a camera mounted at the front, the other with cameras at the front, rear and both sides. They house "FrontVision," which uses a single camera to recognize vehicles ahead, lane markings, road boundaries and traffic signs, and "SurroundVision," which uses four cameras to monitor blind spots around the vehicle.
Inside the demo vehicle, a monitor displayed real-time road conditions along with nearby pedestrians and obstacles, each distinguished by color and tag. With SurroundVision, parking was carried out automatically, with no need to touch the steering wheel, gear shift or brakes. FrontVision has already entered commercial service in overseas markets including India and China, while SurroundVision has secured global mass-production orders and remains under development.
ADAS Adoption Spreads Worldwide, Advancing Camera-Based Autonomous Driving
By integrating and upgrading such camera-based recognition technology, StradVision plans to complete an initial model of MultiVision, which supports autonomous driving at Level 3 and above, starting next year.
"We are currently running joint development projects with multiple global customers," said Kwon Tae-san, StradVision's chief operating officer. "Our goal is to finish first-phase development as early as the middle of next year, then discuss mass-production projects with OEMs to reach Level 3 to Level 4 autonomous driving."
Behind StradVision's push to accelerate is the spread of vehicle safety regulation across countries. India will phase in mandatory advanced driver assistance system features, including lane departure warning systems and blind spot information systems, for new buses and freight trucks starting next year. The United States has also signaled that automatic emergency braking systems will be required on most new passenger cars and light vehicles in 2029, and is tightening performance standards for automotive perception software. StradVision's strategy is to use its road data processing platform to rapidly supply software that meets each country's safety standards and capture global demand ahead of rivals.
From Cattle on the Road to Rear-End Collisions, Synthetic and Simulation Data Fill the Gaps
The foundation for rapidly developing software tailored to differing road environments worldwide is StradVision's "data flywheel." It connects everything from screening and labeling real-world driving data to training, evaluating and improving AI models in a single automated pipeline.
Rare situations that are difficult to capture on actual roads are supplemented with synthetic data. If footage of cattle appearing on an Indian road is needed, three-dimensional assets of cattle are composited into real road video to create training data. Hazardous situations that cannot be deliberately reproduced, such as traffic accidents, are also generated as simulation data for training.
"We will raise the share of synthetic data from about 5% of all training data today to 30% by next year," Kim In-soo, head of StradVision's data innovation center, said that day. The company will also move from next year to commercialize its in-house data pipeline, "SV DataFlow," and offer it to outside customers. That would expand data preprocessing, conversion and labeling technology, used internally as a development tool for autonomous driving software, into a separate SaaS business.
"From the earliest stage of autonomous driving development, we built synthetic data based on real driving data from around the world, and went as far as building a platform that can process it at scale to repeatedly evaluate and improve models — that is what sets us apart," Kwon said. "We will secure more mass-production opportunities by updating software quickly while also verifying safety."
AWS Cloud Handles Large-Scale Data Processing, Training on Traffic Signs From 32 Countries in Two Months
The heavy computing load involved in building the data pipeline is handled by hybrid infrastructure combining the company's own data center with AWS. Routine AI model training runs on its data center in Pohang, while AWS GPU resources are deployed for tasks such as generating large volumes of synthetic data in a short period.
Kwon cited the European Union's traffic sign data project as a representative case. "By running simulations in parallel using AWS infrastructure, we built data on 1,642 types of traffic signs from 32 countries in just two months," he said.
Cooperation with AWS is expected to deepen as AI models such as MultiVision grow in scale. "As AI models get bigger and require more computing resources for training, our reliance on AWS will grow as well," Kwon said.
