Thesis Proposal - Vision-Based Passenger Fall and Abnormal Posture Detection for Autonomous Buses

Autonomous buses are designed to operate without a driver who can continuously observe the passenger compartment and respond to

safety-critical situations. This creates a need for automated systems that can monitor passengers and identify potentially

hazardous situations inside the vehicle.

Recent advances in computer vision and deep learning enable human detection, tracking, pose estimation, and activity recognition.

However, monitoring passengers inside a bus is challenging due to occlusions, limited camera viewpoints, varying illumination,

crowded conditions, and diverse passenger postures.

This thesis aims to develop and evaluate a vision-based system for detecting passenger falls and abnormal postures in autonomous

buses. RGB cameras will be the primary sensing modality, while infrared, thermal, or depth cameras may also be investigated.

• Evaluation of accuracy, false positives/negatives, robustness, and processing latency The work will include a review of

existing approaches, collection and analysis of representative passenger data, development and evaluation of the detection

system, and validation under realistic autonomous-bus scenarios.

Meritorious: Experience with OpenCV, PyTorch, TensorFlow, YOLO, pose estimation, object tracking, activity recognition, ROS2,

video processing, or real-time/edge AI.

Tags: Autonomous Vehicles, Autonomous Bus, Computer Vision · Passenger Safety, Fall Detection , Pose Estimation, Deep Learning,

Object Tracking.

To give you the best possible support during your thesis, we’d like you to be able to come to the office connected to the project

and spend most of your time working from there.

We look forward to receiving your resume, and preferably, a personal letter in which you explain why you want to write your thesis

with Syntronic.