We are developing a multi-modal foundational model for mobile robots called the Kanaria Robotic Model (KRM). KRM is designed to streamline the deployment and operation of mobile robots across diverse environments by enabling them to process multiple types of input data such as voice, images, depth, and sensory information more intelligently. This multi-modal capability allows robots to better understand their surroundings and interact more naturally with humans.

One of KRM’s standout features is its ability to create a simulated environment during deployment. This simulation, which is accessible to staff, allows them to observe the robot’s actions within a virtual setting before they occur in the real world. This feature enhances transparency and safety by providing explainable AI (XAI) insights, showing exactly how the robot will behave.

KRM is also designed to be human-friendly, meaning it can communicate its internal states and reasoning processes to humans while operating. For example, if the robot encounters a problem, such as an obstacle it cannot overcome, KRM can actively ask nearby people for help by clearly explaining what it needs. This ability to engage with humans in real-time not only builds trust but also ensures that the robot can continue to operate effectively in dynamic environments.

There are two ways to utilize KRM:

1. Direct Deployment: KRM can be directly integrated into Kanaria robots to enhance their capabilities.

2. API Licensing: Alternatively, we offer the KRM API to third-party robotics companies. In this scenario, KRM serves as a “teacher” model that supervises and transfers knowledge to a “student” model, which is deployed on third-party robots. This knowledge distillation process allows the student model to inherit advanced capabilities from KRM, enabling it to adapt to specific environments more effectively.