The global market for collaborative manufacturing robots is growing rapidly due to the increasing demand for automation across various industries. According to Global Market Insights, the market size was estimated at around $121.7 million in 2024 and is expected to grow significantly by almost three times by 2034. This growth is attributed to the fact that more and more companies are implementing robots to improve the efficiency and safety of their manufacturing processes.
However, to maximize the accuracy and efficiency of robots, it is necessary to train the AI models that control their behavior.
The concept of industrial robotics
Industrial robotics is a branch of engineering that focuses on creating and applying robotic systems to automate production processes. Robots can perform tasks of varying complexity, from simple parts handling to precision welding and quality control.
The advantage of automated systems is their ability to operate in hazardous conditions.
Automated systems cover the basic needs of:
- Fast assembly and installation of parts.
- Quality control using computer vision allows you to identify product defects quickly.
- Equipment maintenance and routine technical operations.
- Packaging and sorting of products.
The Role of Annotation and Machine Learning
Within the concept of industrial robotics, it is also worth mentioning how data is prepared for machine learning and, in particular, what types of annotations are available:
- Text: keywords, natural language processing.
- Images: bounding boxes, segmentation, cuboid, polygon, skeletal, key points, lane, instance segmentation, bitmask.
- Audio: transcriptions, sound classification.
- Video: bounding boxes, skeletal, key points, lane, instance segmentation, bitmask, 3D point cloud, analysis and recognition of actions, tagging.
- 3D markup: highlighting and tagging point clouds in three-dimensional space with LiDAR sensors.
How does computer vision work in the context of industrial robots? First, the camera captures images, often in several spectra (visible, infrared). Then, the data goes through pre-processing, object classification, and decision-making stages.
High-quality data annotation creates the foundation for industrial robots to “see” and “understand” their environment and recognize text and audio instructions for more flexible control. Such companies as Keymakr use their solutions to annotate training datasets, guaranteeing the speed, quality, and efficiency of training datasets, ensuring the adaptability of robots in any production environment.
Industrial robotics applications
In March 2025, Hyundai Motor Group officially opened Metaplant America (HMGMA) in Ellabell, Georgia, USA. This facility has become one of the most modern plants for producing electric vehicles worldwide, including the IONIQ 5 and IONIQ 9 models. The plant is equipped with advanced technology, including nearly 300 automated guided vehicles (AGVs) and 475 robotic arms, which minimizes human involvement in the production process and increases efficiency.
The plant uses modern solutions, such as autonomous robots with computer vision. To implement such projects, various types of annotations can be used, such as:
- Polygon Markup for precise feature selection;
- Bitmask Annotation for complex parts;
- Bounding Boxes for quick localization of large objects, such as car bodies;
- Keypoint Markup is used to track moving components during assembly.
Integrating cobots into production
In 2024, the BMW Group began piloting the humanoid robot Figure 02 at its Spartanburg, South Carolina, USA plant. California-based Figure developed this robot to perform tasks requiring human dexterity and precision. During the tests, Figure 02 successfully installed metal parts into special fasteners, a key step in assembling a car body.
This implementation is part of the BMW iFACTORY strategy, which aims to increase efficiency, digitalization, and sustainability in production. Using humanoid robots reduces the burden on employees, especially in tasks involving physical activity or repetitive actions.
Universal Robots, a leading manufacturer of collaborative robots, demonstrated the use of deep learning for part detection at IMTS. The main goal of this innovation was to improve the part processing process on industrial lines, where robots must quickly and accurately find and capture various objects. To do this, polygonal annotations were used, which allowed robots to more clearly “see” the contours of complex part shapes.
In 2024, the IEEE/ASME MESA conference presented a study on introducing mobile collaborative robots (cobots) in small and medium-sized enterprises (SMEs) in the Italian automotive industry.
Robots with a sense of touch: a new level of interaction in warehouse operations
Amazon has unveiled its new Vulcan robotic system, which is equipped with a touch sensor. This allows it to efficiently handle about 75% of the products in the company’s warehouses, a significant step forward in robotics development. Vulcan can manipulate objects in confined spaces, such as fabric containers with compartments, which was previously difficult for robots without tactile feedback.
Vulcan’s key features:
- Tactile sensitivity. Vulcan is equipped with a limb with force and torque sensors, which allows it to accurately determine when and with what force to contact objects, preventing them from being damaged.
- AI-driven learning. The robot uses machine learning algorithms to improve its skills, learning from its mistakes and adapting to various situations.
- Collaborative work with people. Vulcan is designed to collaborate with people, performing physically demanding or dangerous tasks, such as accessing upper and lower shelves, which reduces the risk of injuries among workers.
The combination of AI and robotics is opening up new horizons in production processes, with more and more Small and Medium-sized Enterprises (SMEs) beginning to integrate mobile collaborative solutions to increase production flexibility and improve employee working conditions.