Inspiration

Inspired primarily by the love for game design and the desire to collaborate, fascinated by high-speed motorsports like F1 and drag racing, TrackSims was ideated. With the aim to explore the dynamics of multiple autonomous agents under varying conditions, the goal is to simulate realistic race scenarios and extract actionable insights for race management and strategy.

What it does

TrackSims models high-speed agents such as F1 cars, drag cars, jets, and helicopters, each with unique physics profiles and performance attributes. It simulates environments with controllable variables like weather, track conditions, gravity(YES FR! you'll even have different planets as environments ;)), and atmospheric effects. Post-race, the simulator will analyze telemetry data to produce insights on optimal race strategies, timings, and vehicle interactions. You'll be able to configure agents and control the race environment as well, for fun and strategy purposes.

How we are going to build it

We’ll build it in phases: first creating a physics-based simulation core for diverse vehicles with tunable parameters, then adding dynamic environments (weather, gravity, crashes) using Unity or Unreal(Unreal is the preferred engine in such cases as it has better control over physics and real time weather simulation) . A control layer will manage race logic and telemetry collection, which a mostly Python backend will analyze to generate strategic insights. Finally, a simple UI will connect simulation, analytics, and multiple view modes for an interactive, data-driven racing experience.

Challenges we might run into

Building this kind of system will present several real technical hurdles. The first is achieving accurate and consistent physics across very different vehicle types like cars and jets. Handling real-time performance under complex weather and environmental conditions will also be difficult, as both physics and AI processing can strain resources. Multi-agent synchronization, especially during collisions or cooperative behaviors, can lead to instability if not managed carefully. Telemetry render pipeline collection and data pipelines must stay lightweight yet precise, which is difficult when handling large amounts of continuous data. Designing AI that reacts intelligently to physical feedback like traction loss or turbulence will require deep tuning or reinforcement learning. Finally, balancing realism, responsiveness, and cross-platform stability will be a constant challenge throughout development. - For a scalable product , if used in the industry it has got to be accurate to a point where it could be influential to the decisions being made in the actual race.

What we expect to learn

We expect to learn how complex systems behave when physics, data processing, and decision-making come together in real time. Using a game engine, we’ll design and tune dynamic agents, add procedural map generation for diverse environments, and simulate weather conditions like rain, wind, and temperature to see how they affect system behavior. This will help us understand how data turns into insight and how simulation, physics, weather, and AI interact. We’ll gain hands-on experience building intelligent systems that adapt to their surroundings, connecting physics, world design, and machine learning with our broader interest in autonomy and human-machine interaction.