Polarimetric Terra-mechanic Sensor

ICS 496 Capstone Project - UH Manoa S25

Project Team

• Kai Garcia: Hardware Design & Data Pipeline GitHub Profile
• Mairi Yoshioka: Machine Learning & Software GitHub Profile

Figure 1: Neural Network Source. Adapted from [1].

Project Details

Polarimetric Wheel Terramechanic Sensor

End Goal: Deliver a working prototype of a polarimetric terramechanic wheel sensor.

System Components:

1. Polarization camera and light mounted inside the wheel
2. Neural network for data processing

Figure 2: Conceptual diagram of a vision-based terrain traversability system using an in-wheel polarization camera. Adapted from [2].

Weekly Summaries

Week 1:

• Established sponsor and student goals and expectations
• Introduced project details

Week 2:

• Conducted research on polarization and PyTorch

Week 3 (2/5-2/12):

• Set up the camera
• Downloaded the SFP-Wild model

Week 4 (2/13-2/19):

• Developed camera pipeline to NumPy
• Ran SFP-Wild model on example images

About

This project is part of the ICS 496 Capstone Project, focusing on developing an in-wheel polarization camera sensor.

Content sourced from the wheelsensor-ICS496 repository.

Citations

[1] Lei, C., Qi, C., Xie, J., Fan, N., Koltun, V., & Chen, Q. (2022). Shape from Polarization for Complex Scenes in the Wild. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 12632-12641. GitHub Repository

[2] Kalra, A., Taamazyan, V., Rao, S. K., Venkataraman, K., Raskar, R., & Kadambi, A. (2020). Deep polarization cues for transparent object segmentation. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 8602–8611.