Page 385 - AI for Good Innovate for Impact

AI for Good Innovate for Impact

2�2 Benefits of the Use Case

This solution addresses the shortage of expert teaching staff by enabling immersive, remote
knowledge-sharing experiences that can be delivered with minimal infrastructure. It facilitates
realistic and interactive live teaching across geographically remote or rural areas, enhancing 4.3 - 5G
student engagement and improving access to quality education. By narrowing the gap in
educational opportunities across different regions, the use case contributes to greater equity
in learning outcomes.

Additionally, the platform offers strong potential for transforming professional training and
upskilling programs. It enables trainers to deliver scalable, high-quality sessions remotely,
fostering equal opportunities for workforce development and expanding access to economic
growth. By supporting wider reach and efficiency in knowledge delivery, the use case helps
create a more inclusive and adaptable training ecosystem.

2�3 Future Work

Following actions are planned:

• The initial implementation lacks the desired real-time performance. It has been observed
that we need to do more improvement in the human motion prediction algorithm. Some
algorithmic enhancement is planned.
• At present the semantic encoding part is happening at the endpoint itself. This needs to
be transferred to an edge computing facility. This will add some more complexity as the
video needs to be transmitted to the Edge infrastructure. We are considering whether
splitting the AI operation between the Edge and the endpoint may help.
• A standardization of the data format for the semantic exchange is to be devised for
transferring the human motion parameters.
• The application-layer protocol is to be designed with proper loss-resilience and recovery
mechanisms.
• The audio of the teacher is to be integrated with the semantic body description and a
standard to be proposed.

Additional resources:

• The human motion prediction model needs to improve for more accurate prediction and
a new data set to be created.
• The network infrastructure with the Edge service to be made available.

3 Use Case Requirements

• REQ-01: It is critical that the Edge computing services with sufficient GPU are available
near the endpoints along with the network.
• REQ-02: It is critical that the Edge computing infrastructure must allow the application
service provider to host the necessary AI algorithms as specified.
• REQ-03: It is critical that a cloud-service must be available to maintain the session.
• REQ-04: It is critical that the algorithms for generating the semantic information related
to human motion-capture perform in real-time (approximately in 100ms considering a
framerate of 10 fps)
• REQ-05: It is of added value that the rendering algorithm is adaptive considering whether
the trainees are watching through AR/VR glass or watching through large display screens.

349