F1tenth — Time Trial
Overview
Race an F1-Tenth car around a track using any combination of onboard sensors (LiDAR, camera, IMU) and algorithms. Hardware modifications are not permitted; one car races at a time.
Race Scoring
- Base score: time in seconds to traverse the track three times
- +5 seconds for each incidence of grazing or crashing into a barrier
- +10 seconds each time the car needs to be rescued (car is placed in the center of the lane where it was rescued)
Lowest time wins.
Grading
Focus on completing the race without any rescues in the smallest amount of time.
System Modules
| Module | Description |
|---|---|
| Car State | Virtualizes ROS topics into Python-accessible functions and data |
| Controller | Computes control output given current and previous car states |
| LiDAR/Camera perception | Wall detection and obstacle detection |
| Mapping | Builds a map of the track environment |
| Localization | SLAM or vision-based localization using prior map or live sensing |
| Path planner | Determines the optimal path through the track |
Milestones
| # | Time | Milestone | Description |
|---|---|---|---|
| 1 | First review | Setup | Setup simulator and become familiar with it. |
| 2 | Simulation | Develop and test all software modules in simulation | |
| 3 | Lane following (IRL) | Integrate and test lane following with a single vehicle on the physical track | |
| 4 | Robustness testing | Test under varied conditions (different lane widths, lighting, etc.) | |
| 5 | Mid-semester race | First race at the halfway point of the project | |
| 6 | Final presentation | Final race | Final race at the end of the semester and launch project site |
Bonus Points
- Final race: first and second place earn bonus points
Primary Mentor
James