This AI Lets You Program Robots With Plain English—No Coding Needed

Programming a robot usually requires an engineer who understands complex code. Even drag-and-drop block tools are often too rigid for real-world tasks. APPROVE, a new research project, tackles this by using large language models—the same kind of AI behind ChatGPT—to translate plain English instructions into robot programs automatically. You might type, “Pick up the blue box and place it on the shelf,” and the system works out the steps.

But APPROVE doesn't just let the AI run wild. It shows you what it's about to do as a visual block diagram, similar to puzzle pieces on a screen. You can review each step, make changes, or reject the whole plan before the robot does anything. This human-in-the-loop approach means you stay in control—a big deal when you're trusting a machine to move physical objects.

Another clever feature is memory. Once you approve a block sequence, it's saved in a personal library. The next time you need a similar task, the robot can reuse that proven recipe. Over time, the system builds a collection of reliable actions, saving you from starting from scratch. This makes programming less like coding and more like teaching a careful apprentice.

The research is still early, and it's not yet a product in your local factory or kitchen. But it points to a future where factory workers, small business owners, and even home users could customize robots without hiring a programmer. By combining AI's language skills with a simple visual check, APPROVE shows how to make automation more transparent and trustworthy—one approval at a time.