Electric Vehicle Race Optimization

---

Aim:

The use of autonomous vehicles is increasing, and driverless races are starting to take form. This project is aimed at driverless races where a virtual driver could be trained on a virtual racetrack.
So the goal is to combine an existing car model with a racetrack model, programmatically determine the optimal path for the vehicle and tune the suspension/tires of the car for the best possible lap-time.

Tasks:

• Familiarizing with model-based design (focusing on Simscape Multibody, Vehicle Dynamics Blockset, Automated Driving System Toolbox, Mapping toolbox, optimization toolbox)
• Use circuit geometry to determine optimal racing line for point mass model.
• Integrating existing high-fidelity vehicle model into race simulator.
• Tune the suspension hard points for better lap times.
• Optimizing the lap for power consumption.
• Including support for gaming input devices (Gamepad/Steering wheel).

---

Optimization Methodology:

Global Optimization ToolBox:

• fminunc
• fminsearch
• Simulated Annealing
• Pattern Search

Custom Optimization Algorithms:

• Genetics algorithm
• Simulated Annealing

---

Results:

---