List All Projects List Graduate Projects List Senior Design Projects List REU Projects Other Undergraduate Projects
Project # 97
Resilient Multi-Modal Detection Under Sensor Failure
| Team Lead: | Raul Shakya |
| Other Team Members: | Aavash Kharel |
| Faculty Advisor(s): | Dr Debashri Roy |
| Department: | CSE |
| Project Type: | REU |
| Title: | Resilient Multi-Modal Detection Under Sensor Failure |
| Abstract: | obust environmental perception is the cornerstone of resilient autonomous systems, particularly in scenarios prone to sensor degradation or failure. This research plans to establish a high reliability visual baseline for a multi-modal detection framework. We aim to leverage state of the art object detection architectures like YOLOv8 to engineer a custom deep learning pipeline trained on a large-scale proprietary dataset of environmental footage. The project will involve the curation and annotation of thousands of frames to create a specialized ground truth for distinct subject classes. To ensure the resilience of the model to real-world variability with a large number of data, we plan to implement a rigorous training protocol involving random data partitioning and cross-validation against unseen video sequences. We anticipate the resulting model will demonstrate high precision identification and temporal consistency even when applied to raw high resolution sensor streams outside the training distribution. This work aims to provide the critical visual anchor needed to detect anomalies and correct for failures in future multi-sensor integration phases. |
| Poster: | View Poster #97 (PDF) |
| Judges Scoring: | Submit Evaluation (judges only) |