Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Winning Through Simplicity: Autonomous Car Design for Formula Student

Tobias Friedrich, Marco Müller, Adrian Bauske, Simon Härtl, Johannes Herrmann, David Förster, Tobias Tietze, Sebastian Sartor

TL;DR

The paper presents a self-designed autonomous race car by Elefant Racing for Formula Student Driverless, emphasizing a simple, attachable, and robust platform suitable for small teams. It combines an EKF-based state estimator, CENTERNET-based perception with stereo depth, and FastSLAM 1 localization, all guided by a Model Predictive Control planner with a horizon of $H=40$ solved every 50 ms and a blended kinematic/dynamic vehicle model. A hardcoded track centerline for Acceleration/Skidpad and a cone-gate based centerline discovery for Autocross/Trackdrive enable reliable navigation across known and unknown tracks, while the hardware design (DCU, EBS, wheel-hub motors, and steering actuator) supports future expansion. The work demonstrates that a compact, modular, and deliberately simple autonomous system can achieve competitive performance and practical impact, as evidenced by the engineering design award and multi-discipline functionality.

Abstract

This paper presents the design of an autonomous race car that is self-designed, self-developed, and self-built by the Elefant Racing team at the University of Bayreuth. The system is created to compete in the Formula Student Driverless competition. Its primary focus is on the Acceleration track, a straight 75-meter-long course, and the Skidpad track, which comprises two circles forming an eight. Additionally, it is experimentally capable of competing in the Autocross and Trackdrive events, which feature tracks with previously unknown straights and curves. The paper details the hardware, software and sensor setup employed during the 2020/2021 season. Despite being developed by a small team with limited computer science expertise, the design won the Formula Student East Engineering Design award. Emphasizing simplicity and efficiency, the team employed streamlined techniques to achieve their success.

Winning Through Simplicity: Autonomous Car Design for Formula Student

TL;DR

The paper presents a self-designed autonomous race car by Elefant Racing for Formula Student Driverless, emphasizing a simple, attachable, and robust platform suitable for small teams. It combines an EKF-based state estimator, CENTERNET-based perception with stereo depth, and FastSLAM 1 localization, all guided by a Model Predictive Control planner with a horizon of solved every 50 ms and a blended kinematic/dynamic vehicle model. A hardcoded track centerline for Acceleration/Skidpad and a cone-gate based centerline discovery for Autocross/Trackdrive enable reliable navigation across known and unknown tracks, while the hardware design (DCU, EBS, wheel-hub motors, and steering actuator) supports future expansion. The work demonstrates that a compact, modular, and deliberately simple autonomous system can achieve competitive performance and practical impact, as evidenced by the engineering design award and multi-discipline functionality.

Abstract

This paper presents the design of an autonomous race car that is self-designed, self-developed, and self-built by the Elefant Racing team at the University of Bayreuth. The system is created to compete in the Formula Student Driverless competition. Its primary focus is on the Acceleration track, a straight 75-meter-long course, and the Skidpad track, which comprises two circles forming an eight. Additionally, it is experimentally capable of competing in the Autocross and Trackdrive events, which feature tracks with previously unknown straights and curves. The paper details the hardware, software and sensor setup employed during the 2020/2021 season. Despite being developed by a small team with limited computer science expertise, the design won the Formula Student East Engineering Design award. Emphasizing simplicity and efficiency, the team employed streamlined techniques to achieve their success.
Paper Structure (19 sections, 5 equations, 4 figures, 1 table)

This paper contains 19 sections, 5 equations, 4 figures, 1 table.

Table of Contents

  1. Introduction
  2. Design Constraints and Objectives
  3. Autonomous Hardware
  4. Electric Motors for Acceleration and Deceleration
  5. Emergency Brake System (EBS)
  6. Steering Actuator
  7. Processing Units & further Electronics
  8. Sensors
  9. GNSS, IMU and ground speed velocity sensor
  10. Camera and LiDAR system
  11. Synchronization and Calibration
  12. Testing & Data collection
  13. Software
  14. State Estimation
  15. Perception
  16. ...and 4 more sections

Figures (4)

  • Figure 2: Pneumatic and Hydraulic Circuits of EBS
  • Figure 3: Steering actuator with cross section and placement
  • Figure 4: Test Platform
  • Figure : 1: Driverless Compute Unit (DCU); 2: Emergency Brake System (EBS); 3: GNSS/IMU-system; 4: Electronic Box with LTE router; 5: GNSS antennas; 6: Monocular cameras (near: 125° FOV; far: 50° FOV) in orange and stereo camera (45° FOV) in blue; 7: Ground Speed Sensor