Little did we know at the beginning of this course that our team will win at the International Conference of Robotics and Automation, Philadelphia - 2022!
Our team beat the teams from the best universities around the world and also the three time champion at F1Tenth. Great job guys!
My team at ICRA 2022.
Module I: Reactive Methods
Wall Following:
Disparity Extender:
Artificial Potential Fields:
Race 1
Problem Statement
Loop 2 laps around Levine Hall using a reactive algorithm like wall follow or a follow the gap.
Approach
We used a combination of artificial potential fields and cubic splines to plan a follow the gap approach to go around the Levine Hall lobby.
Race Strategy
We quickly realized that the controller response for the straight strech and the corners had to be starkly different, thus we developed a piecewise PID Control algorithm which would switch between two different tuning paramters based on the straight distance from the f1tenth car.
Best friend gifted me an action cam on my birthday! Finally was able to pull this FPV video off the actual car.WE STOOD 3RD IN THE RACE!
Module II: Map Based Methods
Module III: Race Optimization!
Race 3
Problem Statement
Complete fastest 10 laps at Skirkanich Hall Loop whicle racing an opponent's car.
Approach
We used an MPC based controlled but quickly realised the lack of compute causing the car's action to lag. We quickly switched gears to a simple pure pursuit controller strategy while avoiding opponent using a virtual lane.
Race Strategy
Tuning Tuning Tuning!
Tuned the best trajectory and controller gains to get maximum momentum approach where the car slowly began to drift at corners and turned without braking.
Best friend gifted me an action cam on my birthday! Finally was able to pull this FPV video off the actual car.WE STOOD 1ST IN THE RACE!
