AI & Robotics / 2026
CompletedAEGIS-JD
A humanoid robot that responds to gesture, emotion, pose, and voice.
THE PROJECT
From concept
to implementation.
A unified multimodal control system for the EZ-Robot JD humanoid, developed during a six-week internship at the Educational Robotics Lab, NUST SEECS.
The application brings together computer vision, local speech transcription, conversational AI, object detection, face recognition, and a security gate in one dashboard. A Python-to-ARC bridge connects software decisions to physical robot actions.
Gesture, emotion, and pose modes use an 80% confirmation threshold across the last 10 frames, followed by a cooldown, to reduce accidental commands.
My contribution
Group Leader. Built gesture, emotion, pose, and speech-control modules; co-developed Ask JD; proposed and integrated the face-recognition security gate; led system integration. Object detection and the original face-recognition modules were developed by teammates.