Optimization Meeting Minutes

Meeting overview

The goal of the meeting is to discuss possible strategies of solving the optimization problem of figuring out the car speed depending on the environment and car conditions.

Key attendees: @Prabhav Khera @Vinayak Bector @Medhansh Hinduja

Open action items

Discuss inputs that we wish/can integrate into any possible model for optimization

Discuss possible strategies for computing optimized speed for solar car based on previous research papers and past solar car races

Come up with possible implementations of the selected strategies and then present them for the team to decide upon.

Meeting minutes

[01/14] We discussed 2 possible optimization strategies:

Simulation:
- Simulate a given track’s data (repeatedly), modifying constraints, to find the optimal speed of the car on various parts of the track.
- Based on the strategy presented in this paper: https://www.sciencedirect.com/science/article/abs/pii/S0389430497000611?via%3Dihub or use some sort of existing simulation algorithms as in https://github.com/Rooke/qSolarSim
Machine Learning algorithms
- Linear Regression: The goal here is to use the input data, in the form of latitude, longitude, altitude, temperature, humidity, solar irradiance, State of Charge (SoC), cloud cover etc to figure out the best possible speed for each segment.
  - Main Drawback: Needs a lot of training data to work with along with expected correct speed values, which is very hard to build and can take a long time if we try and implement some sort of Simulation for it.

References

Solar car cruising strategy and its supporting system. JSAE Review, 19(2), 143–149.
Rooke. (n.d.). Rooke/qSolarSim: This project is a simulator and data visualization tool used by the Queen’s University Solar Vehicle Team in the 2005 American Solar Challenge. its purpose is to simulate a solar car’s battery, speed, power etc. for a given stretch of road. GitHub. https://github.com/Rooke/qSolarSim