Sereact recently revealed PickGPT, a novel robotics transformer that merges Large Language Models (LLMs) with Computer Vision. This development paves the way for guiding robots using natural language, allowing anyone, regardless of their background knowledge in robotics, to interact with these machines.
LLMs came under the spotlight with the advent of ChatGPT, and their utility expands significantly when integrated with other technologies. Recognizing this, Sereact designed PickGPT to leverage the power of LLMs, which are similar to the models deployed in ChatGPT, combined with Sereact’s patented computer vision models. As a result, robots gain the ability to process and correlate visual information and natural language, enhancing their perception of the environment and their understanding and execution of instructions.
PickGPT, according to Sereact, is the first software-based robotics solution that does not require programming or training. The implications of this are particularly significant for personnel in warehouses, who can now use voice-based instructions or intuitive user interfaces to control robots instead of engaging in complex programming.
As stated by Ralf Gulde, the Co-Founder and CEO of Sereact, “PickGPT facilitates straightforward communication between humans and robots through voice commands and intuitive interfaces. This simplifies the integration with other warehouse systems and allows employees to instruct the robot in a simple manner – for instance, to define pick points or to blacklist certain items.”
This technology not only minimizes the time required to set up and customize robots, thereby freeing up employees to concentrate on other critical tasks, but it also makes advanced robotics technology accessible to individuals without prior training or knowledge.
Moreover, the flexibility PickGPT provides to robots to adapt to new tasks or changing environmental conditions proves crucial in the dynamic and rapidly evolving work environments typically found in the logistics industry.
The capacity of PickGPT to identify objects and products in real-time based on their visual characteristics further enhances its utility in quality control processes. The technology can automatically analyze product images and match them with items in real-world scenarios, in addition to interpreting product titles, descriptions, and other textual data to automatically identify and assign corresponding products to regions in an image.
In the realm of e-commerce, returns processing often poses significant challenges. PickGPT addresses this issue by automating the process, thus saving time and resources. The technology can discern between packaging materials and products, and can sort items automatically based on a description or product image, eliminating the need for barcode scanning.
The launch of PickGPT marks an important development in the robotics technology landscape. By enabling intuitive operation using natural language, the technology makes AI robotics accessible to individuals without previous knowledge. PickGPT also provides solutions for pressing issues in warehouses, such as improving efficiency in quality control and expediting returns processing.