1 of 9

Self-Driving Vehicular Project

Arya Shetty

Tommy Chu

Aaron

Cruz

Brandon Cheng

Advisors: Ivan Seskar and Jennifer Shane

Sid

Malhotra

Erik Nießen

2 of 9

Project Background

Smart City self-driving

RASCAL (Robotic Autonomous Scale Car for Adaptive Learning)

  • Neural Network

SCAMP (Self-guided Computer Assisted Mecanum Pathfinder)

  • NPC

3 of 9

This Week's Progress

Curvature Interpolation

Calibration

4 of 9

This Week's Progress

Double barrier data

5 of 9

This Week's Progress

Training batches → Less 0 biased

6 of 9

This Week's Progress

Data augmentation blur artifacts during NN normalization

→ Solid fill

7 of 9

This Week's Progress

Goal: Evaluate models more accurately than loss function

Simulated driving with ML model and image skewing

With Heatmap

Simulated View

8 of 9

This Week's Progress

ArUco Detection Improvements!

  1. Refactored code

2. Added functionality for multiple arUcos

9 of 9

Next Week’s Goals

Thank you!

Any Questions?

Expand inputs for Neural Network

City test data

ML simulator evaluation