Artificial Intelligence of Things (AIoT) refer to the integration of AI directly into Internet of Things (IoT) devices, enabling these devices to process data, make decisions, and execute AI algorithms locally without the need for constant cloud connectivity. This allows IoT devices to operate autonomously, reduce latency, and improve privacy by keeping sensitive data on the robotic device itself. The evolution from cloud-based AIoT to on-device AIoT is being driven by several trends in technology, user needs, and infrastructure limitations. One of the main reasons for moving AI processing from the cloud to the device is the need to reduce latency. The trend from cloud-based AIoT to on-device AIoT is characterized by the need for low latency, enhanced security, energy efficiency, and the availability of more advanced hardware that can handle AI workloads locally. Due to the complexity of integrating AI processing, IoT devices, and diverse application environments, standardization of on-device AIoT poses several challenges. Lack of interoperability, diverse hardware capabilities and heterogeneous on-device computing environments are key challenges and opportunity for standardizing on-device AIoT framework. In particular, on-device AIoT framework for robot services is growing rapidly with the increasing adoption of intelligent robots in various industries. On-device AIoT framework for robot services need to support for seamless integration with robotic tasks and real-time responsiveness, which is critical for the simultaneous execution of robotic operations and intelligence. Therefore, on-device AIoT framework for robot services will enable a more robust and efficient robot ecosystem, benefiting the AI and IoT robot industry and consumers.