News and Announcements
[Press Vision] THOTH Inc. Enhances On-Site Efficiency and Safety with the Robot AI Solution 'RAAPS
2023-07-20
Realizing Automated Processes for Waste Battery Recycling and Logistics Fulfillment
With the advancement of deep learning-based AI solutions and robotic technologies, automation is being applied to increasingly complex and high-level processes. This enables workers to perform tasks more safely and efficiently. THOTH Co., Ltd. (hereafter referred to as THOTH) has developed the Robot AI-based Autonomous Programming Solution (RAAPS), which allows robots to autonomously learn and program using techniques such as imitation learning, inverse reinforcement learning, and reinforcement learning. The company is actively deploying this solution in the expanding secondary battery recycling industry.

The name THOTH, inspired by the ancient Egyptian god known for creating time and calendars, inventing hieroglyphics, and overseeing the development of science, knowledge, and wisdom, represents the company’s commitment to creating safer, more efficient, and more convenient automated environments through the fusion of AI and robotics.
The company uses its RAAPS technology to automate processes in waste battery recycling and logistics fulfillment. This solution enables robots to independently gather and learn the necessary data after observing demonstrations by field workers. The robots then generate programs to automate tasks and execute them with hardware integration.
THOTH CEO Lee Sang-hyoung stated, “Automation has evolved from simple repetition to more advanced processes thanks to deep learning. Robots can now learn from on-site experiences, understand entire processes, and even execute them autonomously. By developing RAAPS, THOTH has enabled robots to understand field conditions and program themselves. We are committed to applying this technology across various industries.”
Enhancing Learning Capabilities for Maximum Efficiency
Deep learning-based imitation learning and reinforcement learning are primarily divided into direct teaching and task observation methods. Direct teaching involves field workers manually teaching processes and components, allowing the robots to simultaneously learn images and control signals. Task observation, on the other hand, enables robots to generate and refine task plans by observing the actions and procedures of field workers in image form, without explicit teaching.
When THOTH receives a request for process automation, it begins with an on-site investigation (or the analysis of photos and videos) to understand the current state. The team then designs a 3D model of the automation process, conducts physical simulations to incorporate client feedback, and refines the solution before implementation.
CEO Lee highlighted that this comprehensive approach offers numerous advantages, such as enhancing client trust, minimizing trial-and-error, and reducing costs and time through simulation-generated learning data.

Currently, THOTH places significant value on simulation-based learning data, allocating 90% to logistics simulation and 10% to real-world data. Lee explained, “Simulations not only reduce learning time but also equip robots with the ability to handle rare or complex exceptions. THOTH is developing solutions where robots collect data on-site, upload it to our servers, and receive updated learning outcomes for adaptive operations.”
RAAPS excels in detailed performance metrics such as programming automation rates, deep learning training times, robot inference/operation speeds, and precision improvements. These strengths translate into significant economic benefits. In the first half of this year, THOTH generated approximately KRW 1 billion in revenue through logistics palletizing and packaging automation, as well as gyro block polishing and inspection automation. The company aims to develop solutions generating an additional KRW 2 billion in revenue in the second half.
Advancing the Recycling Business
As the EV market grows, the secondary battery industry is expanding to include raw materials, cells, modules, packs, and used battery recycling. THOTH is advancing its waste battery recycling efforts by automating processes like inspection, diagnosis, disassembly, and final checks, which are traditionally labor-intensive. This robot-based solution aligns with the company’s philosophy of enhancing efficiency and safety through AI and robotic technologies.
If successfully implemented, this solution could address safety risks such as electric shocks and musculoskeletal disorders associated with manual labor. THOTH aims to automate the disassembly of 40,000 waste batteries annually, processing 50 batteries daily on a single line. By leveraging RAAPS and its advanced learning techniques, the company addresses challenges such as unpredictable battery conditions, compatibility across vehicle models, and precise disassembly of flexible components.

THOTH is also constructing a smart factory in Seongsu-dong, Seoul, designed for fully automated EV waste battery diagnosis and disassembly, as well as logistics and agricultural packaging automation. This approximately 330 square meters facility, set to be completed in late 2023, will serve as both a smart factory and a showroom for showcasing THOTH’s latest AI and robotic technologies.
Aiming for Market Leadership
THOTH plans to raise Series A funding in the second half of this year to establish its position in the waste battery recycling market. Once the funding is secured, the company intends to form partnerships with local governments and related enterprises to activate battery supply chains. THOTH also aims to supply automated process lines to large corporations, expand its capacity, and establish regional centers for global market entry.
CEO Lee concluded, “As the secondary battery industry continues to grow, we will maximize the technological value of our automation solutions to support its advancement.”
Yongjun Kim, Reporter
출처 : [Press Vision] Enhances On-Site Efficiency and Safety with the Robot AI Solution 'RAAPS' - robotzine.co.kr