Robotics Revealed — localization for autonomous driving
Student researcher, UCSB Summer Research Academies
Over the summer I joined the UCSB Summer Research Academies in the Robotics Revealed track. My work centred on localization for autonomous driving: I reviewed 10+ research papers and compared the state-estimation principles and computational complexity of the Extended Kalman Filter (EKF), the Unscented Kalman Filter (UKF) and the Particle Filter (PF).
I then ran localization simulations in MATLAB, setting up seven types of test scenario — including GNSS signal interruptions, severe weather and different sensor noises — to evaluate localization error, robustness and computational overhead. With two teammates I analysed the results and found that the UKF offered the most balanced overall performance while the PF carried a higher computational cost; I took the lead on the 10-page technical report and the final research presentation. Alongside the project I took university-level courses in rigid-body dynamics, coordinate transformations and robotics control, earning an A+ grade.
What the placement involved
- Literature review. 10+ papers on localization, comparing EKF, UKF and PF on principle and complexity.
- Simulation. Seven test scenarios in MATLAB, including GNSS interruption, bad weather and sensor noise.
- Evaluation. Localization error, robustness and computational overhead compared across the three filters.
- Reporting. A 10-page technical report and the final presentation, led by me with two teammates.
- Coursework. Rigid-body dynamics, coordinate transformations and robotics control — A+.
10+
Papers reviewed
7
Test scenarios
3
Filters compared
10
Page report