LITEON Technology (LITEON) is actively deploying new Artificial Intelligence (AI)-assisted vision software technologies enabling a wide variety of features for improved road safety and driver convenience. At AI EXPO Taiwan 2025, organized by DIGITIMES, LITEON showcases a series of breakthrough innovations leveraging existing vehicle hardware and optimized AI algorithms to equip vehicles with software that upgrades driving safety. Brady Jung, Associate Vice President of LITEON Core Competence Technology Research Center, describes the new technology strategy focused on the safe driving sector in this interview.
This research and development organization aims to develop advanced technologies over the next 5 to 10 years, targeting software development to sharpen LITEON's long-term competitiveness. Simultaneously, the technology development team is looking for promising startups for potential investments, pushing the boundaries of what's possible and shaping the future of technology. The following vision software demos assist drivers in detecting potentially dangerous conditions, advancing road safety.
Monitor driver attentiveness to detect 'See You Unseen' and enhance worry-free driving
The first technology monitors driver attentiveness and driving behavior to identify potentially dangerous driving patterns. This vision software leverages DMS (Driver Monitoring System) cameras mounted on the upper windshield area or near the steering wheel to estimate a driver's gaze direction. Jung provides typical examples of distracted driving: a driver makes a right turn with clear signaling but turns his head to the left because he notices something interesting. This technology can detect where drivers are looking by analyzing their head movements and providing real-time safety warnings to prevent potentially dangerous situations. This vision software benefits drivers by monitoring their condition and evaluating how safely they are handling the vehicle.
Panoramic multi-object tracking to determine the movements of objects around vehicle
The second demonstration is the multi-object tracking technology, such as detecting pedestrian movements or other approaching vehicles, to predict possible changes in driving situations on the road. This software uses a combination of AI algorithms to develop a comprehensive understanding of the driving environment, enabling better utilization of the vehicle's camera systems. One of the unique features of this software is consolidating data from multiple cameras into a single, coherent perception result of identified objects. By leveraging surround-view multi-object tracking technology and processing massive image data from multiple cameras, the software can rapidly recognize people and other objects around the vehicle and even predict their future trajectories.
Simulate human vision to detect 3D objects for mitigating potential accidents
Inspired by human binocular vision, the third demo aims to track 3D objects by identifying object depth using two cameras. The camera arrangement can consist of one camera at the front of the car and another on either side, creating a viewing angle difference similar to human binocular perception. This vision software detects the depth of objects. Jung cites a common scenario in which ordinary vehicles struggle to accurately determine the actual location and depth of sidewalk curbs when parking on the roadside. This software helps prevent common mistakes that cause tire and rim damage from hitting curbs. The benefit of this software is improved accuracy in perceiving the 3D shapes of objects, such as road curbs in the above example.
Subtle facial physiological reaction capturing and understanding to predict risks
The fourth technology integrates DMS cameras to capture drivers' facial physiological signs while driving. Even under challenging lighting conditions, this technology can accurately detect subtle facial fatigue signals across wide variations in brightness and diverse driving environments. This vision software utilizes a near-infrared (NIR) camera to capture subtle physiological signs, such as pulse fluctuations on the driver's face or neck. In collaboration with two top universities and their medical schools in Taiwan, several ongoing research projects aim to deepen the interpretation of drivers' physiological signs. It is especially crucial to analyze rapid pulse changes, as understanding these can enhance driving safety and contribute to the development of early risk warning systems.
LITEON explores the crucial role of middleware suppliers with AI vision software
Behind these technologies lies the core strength that LITEON has built up over the years in the automotive market. With expertise in hardware optical devices and cameras, LITEON possesses unique advantages in developing automotive vision software. Nowadays, modern cars are typically equipped with more than 10 built-in cameras as standard features. The massive amount of image data and video streams captured from a vehicle's cameras provides fertile ground for disruptive innovations, unlocking emerging revenue opportunities in the automotive industry. Leveraging powerful AI algorithms, LITEON develops vision software stacks and actively plays the role of automotive middleware provider. LITEON is aggressively selling a business model for vision system integration in the automotive sector.
Jung emphasized several unique strengths that enable LITEON to explore the automotive vision software business. First, LITEON's vision perception technologies offer significantly broader coverage, simultaneously addressing human drivers and the surrounding vehicle environment. Jung refers to this approach as "Human-Vehicle Integration," highlighting the company's core technologies and know-how in the integrated detection of people and vehicles. Second, by utilizing existing Electronic Control Units (ECUs) and vehicle cameras, LITEON implements specialized "Camera-Independent" technology to fine-tune image quality across different car models. This ensures consistent image quality performance, maintaining reliable detection and monitoring capabilities. Third, through its extensive history of collaboration with global brands and automotive Tier 1 suppliers, LITEON has established long-term partnerships and works closely with supply chains, accelerating new product development and significantly reducing customers' time-to-market.
Furthermore, to maximize vision software performance across various ECU systems and vehicle models, LITEON continues to invest heavily in developing lightweight AI algorithms. These optimized algorithms drive computing power efficiency, delivering higher computational performance while being less resource-intensive and more power-efficient across mainstream ECU platforms. This design aligns with long-term automotive industry trends toward sustainable growth, innovative solutions, and meeting consumer and regulatory demands.
As automotive applications evolve, the focus remains on minimizing environmental impact and enhancing driving safety and convenience. In the upcoming phase, LITEON is engaging with global automotive brands and Tier 1s to demonstrate vision software through linking and displaying on the infotainment screen. Additional procedures include preparing a physical demo vehicle for real-time testing and verification on the customer side. Through these efforts, LITEON is actively stepping into the safe-driving software market, expanding its software solutions by leveraging core competencies and incorporating AI-powered features to identify another significant growth engine in the future. For further information about LITEON's solutions, please visit www.liteon.com or view the introduction video titled "See Your Unseen. Worry-Free Driving." here.
Article edited by Jack Wu
