A research team led by Professor Kyungjoon Park from the Department of Electrical Engineering and Computer Science at Daegu Gyeongbuk Institute of Science & Technology (DGIST; President Kunwoo Lee) has developed autonomous driving software that enables low-cost sensors to detect transparent obstacles—such as glass walls—offering a practical alternative to expensive high-performance sensors.
Glass Wall Detection Without the Use of Expensive Sensors. Image Credit: Daegu Gyeongbuk Institute of Science & Technology
The technology can be applied to existing robots without additional hardware, maintaining detection performance on par with more costly equipment.
Autonomous driving robots often rely on LiDAR sensors to perceive their surroundings. These sensors act like “laser eyes,” measuring distances and identifying structures by projecting light and timing its return. However, affordable LiDAR sensors typically fail to detect transparent materials, mistaking them for empty space—an issue that can lead to collisions. While high-resolution ultrasonic LiDAR sensors or camera-based systems can solve this problem, they significantly raise the system's complexity and cost, often by hundreds of thousands to millions of won.
To address this challenge, Professor Park’s team developed a software-based solution called Probabilistic Incremental Navigation-based Mapping (PINMAP). Instead of requiring new hardware, PINMAP enhances the capabilities of existing low-cost LiDAR sensors by collecting rare point data that these sensors occasionally detect and using it to probabilistically infer the presence of transparent barriers over time.
Built on widely used open-source platforms like Cartographer and Nav2 within the ROS 2 ecosystem, PINMAP integrates easily into current systems. It improves how existing sensors interpret data, removing the need for expensive upgrades while significantly boosting detection accuracy.
In real-world tests at DGIST, PINMAP achieved a 96.77 % accuracy rate in detecting glass walls, far surpassing the near-zero detection rate of traditional systems using the same sensors. This performance leap, driven entirely by software, underscores the potential for cost-effective improvements in autonomous navigation.
PINMAP flips the conventional wisdom that hardware performance equals system performance and proposes a new standard whereby software can improve sensor capabilities. This study shows that ensuring stable autonomous driving is possible without relying on high-performance equipment.
Kyungjoon Park, Professor, Department of Electrical Engineering and Computer Science, Daegu Gyeongbuk Institute of Science & Technology
The algorithm offers substantial cost savings by delivering detection capabilities comparable to high-end LiDAR systems at a fraction of the price. It holds promise for reducing collisions in indoor environments like hospitals, airports, malls, and warehouses—key areas for the broader rollout of service robots.
Journal Reference:
Chae, J., et al. (2025) PINMAP: A Cost-efficient Algorithm for Glass Detection and Mapping Using Low-cost 2D LiDAR. IEEE Transactions on Instrumentation and Measurement. doi.org/10.1109/tim.2025.3566826.