06/26/21 - Esteban Strategy Team

These are notes typed up during a meeting with the Esteban Strategy team. We reached out to them to discuss team strategies and how their models work and were willing to have a meeting to talk about how our approach was and compare it to theirs.

Esteban Attendees

• Etienne Bedard - mech lead, strategy lead

• Jeremie Riviere - strategy member

Meeting Goals

• Compare approaches

• Discuss gaps in our strategy model

Questions

• methods of evaluating car state

• driver profile? something to consider/have they considered

• modelling the car over the course of the race, deterioration, etc

Esteban Strategy

• very internal data heavy

• route → 12-15 points

  • solar forecast for points

  • "Fat table" how race should go for each second/point

  • known forecast, elevation/speed etc, what expected battery state is at a given time

  • Split by stage

  • checks if enough time to do something in a day

  • number of loops, energy put in via charger, target speed are the variables

  • speed → assumes ASC min whole time

  • outputs score, repeats for slightly different strategy (ex: 1 less minute of charging, 1 less loop)

• can be started from any given point in a race, takes into account what has already happened

• ignores speed? assumes 65 in most cases

  • interesting since we rely heavily on speed calculations

• does some driver profiling

  • energy consumption

• adapts to previous car performance, can tell them to go faster if car was slower before?

Thoughts from Esteban

• micromanaging speed is not the way, may not be worth it

• loops weren't hard to implement, can decide whether to do or not

  • checks battery, is enough to do loop if not then not worth

  • could use that time to charge instead?

• pilot comparison: speed vs power consumption

  • test on a track

• willing to share carbon fiber knowledge

• if we have to micromanage something: micromanage battery and SOC