What if a robot could understand your instructions like a colleague? KOMPAI Delivery is exploring the integration of an embedded voice assistant to simplify medical logistics and support all users — even beginners.
The Problem: A Useful but Sometimes Intimidating Robot
In clinics and hospitals, the KOMPAI Delivery robot transports medication and medical supplies between the pharmacy and various departments. It automates repetitive tasks and reduces the workload for staff.
But in practice, users frequently change — temporary staff, students, replacements. Not everyone knows how to use the robot:
- How do you launch a delivery?
- How do you confirm it?
- What should you do if something goes wrong?
Training every new user is expensive and time-consuming.
As a result, some hesitate to use the robot, or only call on it when someone nearby already knows how to operate it.
Our Idea: An Embedded Voice Assistant
To remove this barrier, we’re working on a concept: integrating an embedded AI capable of guiding users through voice interactions.
Concretely:
- The robot could understand simple natural language instructions;
- The user could speak directly to the robot or use a small touchscreen;
- The robot would respond clearly and appropriately, asking questions if needed.
The goal?
To enable anyone to use the robot without specific training.
How It Works
The embedded AI would have three main functions:
- Understand natural language instructions or questions (“Retrieve the medication from the robot’s cabinet,” “Send medication back to the pharmacy”);
- Guide the user step by step (“Hello! Where should I deliver?”);
- Confirm or cancel actions (“I’ve received your request,” “I’ve just canceled the delivery”).
To ensure data privacy and autonomy in case of limited connectivity, both speech recognition and language understanding would ideally be processed locally on the robot.
A Concrete Scenario
Example interaction:
Nurse: “Hello, I need to deliver this medication.”
Robot: “Hello! Could you tell me which department?”
Nurse: “Gastroenterology.”
Robot: “Thank you! I’ll take care of it right away.”
This short, natural exchange allows untrained users to operate the robot easily, without pressing complex buttons or remembering procedures.
Technical and Ethical Challenges
Implementing this interface is no small task. Key challenges include:
- Reliable speech recognition in noisy environments;
- Understanding medical domain-specific language;
- Handling different languages and accents;
- Ensuring privacy and security compliance.
These are part of our roadmap: we want a robot that helps without adding risk or complexity.
Expected Benefits
- Easier adoption by occasional or temporary users
- Fewer errors or blockages thanks to contextual guidance
- Time savings for training staff
- Better integration of the robot as a “colleague” rather than just a machine
Why This Is a Relevant Use Case for Neurokit
This project perfectly illustrates Neurokit’s ambition: developing embedded AI that adapts to both the user and the context, with real impact in the field.
Integrating an intelligent voice assistant into KOMPAI Delivery targets a critical and concrete use case: hospital logistics. It shows how embedded AI can not only automate tasks but also enhance human experience and the quality of care.
KOMPAI Robotics is pursuing this development and relies on the Neurokit ecosystem to share, test, and refine these approaches.